中

CIVITAI / PUBLIC LIBRARY

AI-KSK Civitai Models and Workflows

Search AI-KSK public Civitai models, LoRAs, and workflows with versions, base models, and source links.

385

public entries

23

base model families

60732

total downloads

Workflows 2025-12-14

Wan-Move latent trajectory-guided video generation

This workflow introduces the Wan-Move model , using latent trajectory guidance for video generation. It allows precise control over the movement in video creation by defining trajectory paths in a 3D space. This model is SOTA-level in its accuracy and fine-grained control, ensuring that even when movements are not perfect, the results remain consistent and functional. The system includes tools for adjusting movement speed and controlling interpolation between points on the path. You can create a smooth video sequence by inputting trajectory coordinates, which dictate the motion's flow, such as speed, acceleration, and direction. The model can handle complex animations like characters following a path or creating abstract visual effects (e.g., spinning planets or characters moving through space). The trajectory control also extends to scenes requiring extended video generation , enabling infinite loops or precise adjustments to video length, without losing detail or coherence. The model is particularly well-suited for abstract video creation where traditional prompts may not suffice, offering an alternative method of motion control for artistic and cinematic visuals. 🎥 YouTube Video Tutorial Want to know what this workflow actually does and how to start fast? This video explains what the tool is, how to launch the workflow instantly, and shares my core design logic — no local setup, no complicated environment. Everything starts directly on RunningHub, so you can experience it in action first. 👉 YouTube Tutorial： https://youtu.be/xNZBU2seFLU Before you begin, I recommend watching the video thoroughly — getting the full context helps you understand the tool faster and avoid common detours. ⚙️ RunningHub Workflow Try the workflow online right now — no installation required. 👉 Workflow： https://www.runninghub.ai/post/1998898965116542977/?inviteCode=rh-v1111 If the results meet your expectations, you can later deploy it locally for customization. 🎁 Fan Benefits： Register to get 1000 points + daily login 100 points — enjoy 4090 performance and 48 GB super power! 📺 Bilibili Updates (Mainland China & Asia-Pacific) If you’re in the Asia-Pacific region, you can watch the video below to see the workflow demonstration and creative breakdown. 📺 Bilibili Video： https://www.bilibili.com/video/BV1NqmgBREqZ/ ☕ Support Me on Ko-fi If you find m

Wan Video 14B i2v 480p 114 downloads
View public details
Workflows 2025-12-09

Wan-Animate video stylization

This workflow extends Wan-Animate into a reliable video stylization pipeline by combining a style-defining reference image with pose information extracted from the input video. Because the model anchors its appearance to the first frame, any stylized reference—illustration, anime, or textured art—propagates consistently across the entire sequence. The updated pose-alignment module allows the system to match the motion video to the reference image or match the reference to the motion, keeping proportions stable even when the character design is unusual or non-human. This alignment is especially helpful when the reference image is slightly expanded to match the video’s aspect ratio, producing cleaner and more coherent motion guidance. In practical use, the workflow tracks the original pose video faithfully while preserving the visual identity set by the reference. Wan-Animate handles single-person scenes with high stability, and SteadyDancer-style behavior ensures that the model maintains style coherence without drifting. For multi-character or complex scenes, traditional OpenPose detection remains effective, but the updated alignment module improves stability whenever a single subject is involved. The result is a smooth, style-consistent animation that retains the full artistic look of the reference while accurately following the motion of the input video. 🎥 YouTube Video Tutorial Want to know what this workflow actually does and how to start fast? This video explains what the tool is, how to launch the workflow instantly, and shares my core design logic — no local setup, no complicated environment. Everything starts directly on RunningHub, so you can experience it in action first. 👉 YouTube Tutorial： https://youtu.be/Yrq8WiTPnlk Before you begin, I recommend watching the video thoroughly — getting the full context helps you understand the tool faster and avoid common detours. ⚙️ RunningHub Workflow Try the workflow online right now — no installation required. 👉 Workflow： https://www.runninghub.ai/post/1998364370491047938/?inviteCode=rh-v1111 If the results meet your expectations, you can later deploy it locally for customization. 🎁 Fan Benefits： Register to get 1000 points + daily login 100 points — enjoy 4090 performance and 48 GB super power! 📺 Bilibili Updates (Mainland China & Asia-Pacific) If you’re in the Asia-Pacific region, y

Wan Video 14B i2v 720p 137 downloads
View public details
Workflows 2025-12-09

kj new Pose Detection + Wan-Animate

This workflow upgrades Wan-Animate by inserting a new pose-detection module that aligns the motion video with the reference image before animation begins. That alignment step eliminates scale mismatches and keeps proportions stable, so stylized or exaggerated characters transfer cleanly into motion. Because Wan-Animate treats the first frame as the style anchor, the visual look from that single reference image is preserved across the entire sequence with much stronger consistency than traditional video style-transfer. The workflow works equally well for unconventional proportions—chibi, monsters, or distorted silhouettes—because it can align the pose to the reference or the reference to the pose depending on the scenario. In practice, expanding the reference image slightly to match the incoming pose video improves stability, and the updated detector produces cleaner skeletons that reduce jitter in the final animation. For multi-person scenes, open-pose detection still performs reliably, but single-subject sequences benefit the most from the new alignment. Combined with Wan-Animate’s ability to propagate the first-frame style throughout the video, the workflow delivers steady identity retention, smooth motion, and visually cohesive stylization, even when the input video has complex or exaggerated movements. 🎥 YouTube Video Tutorial Want to know what this workflow actually does and how to start fast? This video explains what the tool is, how to launch the workflow instantly, and shares my core design logic — no local setup, no complicated environment. Everything starts directly on RunningHub, so you can experience it in action first. 👉 YouTube Tutorial： https://youtu.be/Yrq8WiTPnlk Before you begin, I recommend watching the video thoroughly — getting the full context helps you understand the tool faster and avoid common detours. ⚙️ RunningHub Workflow Try the workflow online right now — no installation required. 👉 Workflow： https://www.runninghub.ai/post/1998289350926458882/?inviteCode=rh-v1111 If the results meet your expectations, you can later deploy it locally for customization. 🎁 Fan Benefits： Register to get 1000 points + daily login 100 points — enjoy 4090 performance and 48 GB super power! 📺 Bilibili Updates (Mainland China & Asia-Pacific) If you’re in the Asia-Pacific region, you can watch the video below to see the workflow d

Wan Video 14B i2v 720p 72 downloads
View public details
Workflows 2025-12-09

kj new Pose Detection + SteadyDancer

This workflow updates SteadyDancer with a new pose-detection module that aligns the skeleton video directly to the reference image before animation begins. This alignment prevents the usual scale and proportion mismatches that occur when the pose video and the reference have different aspect ratios. Once aligned, SteadyDancer treats the first frame as the visual anchor, preserving identity and stylization through the entire sequence. Because Wan Animate can optionally align the reference to the pose or the pose to the reference, the workflow adapts well to exaggerated designs, chibi proportions, monsters, or any non-standard character shapes. In use, the updated detector produces cleaner, more stable poses, especially when the reference is slightly expanded to match the skeleton’s width and height. This improves downstream motion quality without needing manual fixes. For style-transfer video generation, the model uses only the first frame’s appearance, applying that look across the whole motion clip with strong consistency, even for multi-character or unusual body structures. While open-pose detection remains useful for crowded scenes, SteadyDancer still handles complex or distorted silhouettes gracefully. Overall, this update makes pose-driven animation more predictable, more identity-stable, and more suitable for stylized or unconventional characters. 🎥 YouTube Video Tutorial Want to know what this workflow actually does and how to start fast? This video explains what the tool is, how to launch the workflow instantly, and shares my core design logic — no local setup, no complicated environment. Everything starts directly on RunningHub, so you can experience it in action first. 👉 YouTube Tutorial： https://youtu.be/Yrq8WiTPnlk Before you begin, I recommend watching the video thoroughly — getting the full context helps you understand the tool faster and avoid common detours. ⚙️ RunningHub Workflow Try the workflow online right now — no installation required. 👉 Workflow： https://www.runninghub.ai/post/1998291780594491394/?inviteCode=rh-v1111 If the results meet your expectations, you can later deploy it locally for customization. 🎁 Fan Benefits： Register to get 1000 points + daily login 100 points — enjoy 4090 performance and 48 GB super power! 📺 Bilibili Updates (Mainland China & Asia-Pacific) If you’re in the Asia-Pacific region, you

Wan Video 14B i2v 720p 51 downloads
View public details
Z-Image + Sigma Split dual-stage rendering
Workflows 2025-12-08

Z-Image + Sigma Split dual-stage rendering

This workflow splits Z-Image’s generation into a high-noise structural pass and a low-noise detail pass using Sigma separation. The first stage focuses on shape and control—such as pose, depth, or lines—while the second stage removes ControlNet influence and activates the base model and LoRA more effectively. This two-phase rendering restores the aesthetic richness that ControlNet usually suppresses, giving a cleaner, more expressive image that still follows the structural guide. In practice, the low-noise pass is where LoRA expression, fine texture, and overall visual appeal return. A detail-enhancing module can be inserted here to boost clarity without harming structure. After refinement, facial detailing and tiled upscaling provide a final high-resolution result that stays consistent with the guided pose while achieving a more polished, character-accurate appearance. 🎥 YouTube Video Tutorial Want to know what this workflow actually does and how to start fast? This video explains what the tool is, how to launch the workflow instantly, and shares my core design logic — no local setup, no complicated environment. Everything starts directly on RunningHub, so you can experience it in action first. 👉 YouTube Tutorial： https://youtu.be/sgG7PizsyDo Before you begin, I recommend watching the video thoroughly — getting the full context helps you understand the tool faster and avoid common detours. ⚙️ RunningHub Workflow Try the workflow online right now — no installation required. 👉 Workflow： https://www.runninghub.ai/post/1997541959206125569/?inviteCode=rh-v1111 If the results meet your expectations, you can later deploy it locally for customization. 🎁 Fan Benefits： Register to get 1000 points + daily login 100 points — enjoy 4090 performance and 48 GB super power! 📺 Bilibili Updates (Mainland China & Asia-Pacific) If you’re in the Asia-Pacific region, you can watch the video below to see the workflow demonstration and creative breakdown. 📺 Bilibili Video： https://www.bilibili.com/video/BV1dAmPBXEaR/ ☕ Support Me on Ko-fi If you find my content helpful and want to support future creations, you can buy me a coffee ☕。 Every bit of support helps me keep creating — just like a spark that can ignite a blazing flame。 👉 Ko-fi： https://ko-fi.com/aiksk 💼 Business Contact For collaboration or inquiries, please contact aiksk95 on WeCha

ZImageTurbo 101 downloads
View public details
Z-Image-ControlNet second-order Refiner + Pose / Depth / Line
Workflows 2025-12-05

Z-Image-ControlNet second-order Refiner + Pose / Depth / Line

This workflow uses the Z-Image model together with ControlNet to achieve reliable control over pose skeletons, depth information, and line-based structure. Z-Image’s fast inference makes it responsive to these conditioning signals, producing an initial output that follows the intended structure while still carrying the model’s vivid rendering style. However, ControlNet naturally reduces aesthetic richness, so the workflow incorporates a second refinement stage: the generated image is re-encoded into latent space and passed through another diffusion round. This latent-space pass restores color depth, detail, and visual cohesion more effectively than pixel-space upscaling, giving the final image a more polished look. In practice, switching to SD-Pose improves skeleton accuracy when the automatically detected pose is incomplete, and adjusting preprocessing strength lets you fine-tune how strongly the structure affects the result. Line control and depth control both behave predictably, though stylized cases may still require refinement to maintain visual appeal. LoRA models can be loaded, but ControlNet will weaken their influence, so the workflow favors simpler stylistic adjustments. Despite its early stage of development, Z-Image already shows strong potential—its image-to-image quality is striking, and the two-stage approach offers a balanced combination of structural control and aesthetic refinement. 🎥 YouTube Video Tutorial Want to know what this workflow actually does and how to start fast? This video explains what the tool is, how to launch the workflow instantly, and shares my core design logic — no local setup, no complicated environment. Everything starts directly on RunningHub, so you can experience it in action first. 👉 YouTube Tutorial： https://youtu.be/7GtFnr8ixnM Before you begin, I recommend watching the video thoroughly — getting the full context helps you understand the tool faster and avoid common detours. ⚙️ RunningHub Workflow Try the workflow online right now — no installation required. 👉 Workflow： https://www.runninghub.ai/post/1996508277796126722/?inviteCode=rh-v1111 If the results meet your expectations, you can later deploy it locally for customization. 🎁 Fan Benefits： Register to get 1000 points + daily login 100 points — enjoy 4090 performance and 48 GB super power! 📺 Bilibili Updates (Mainland China & Asia-Pac

ZImageTurbo 764 downloads
View public details
Z-Image-ControlNet + SDPOSE + second-order Refiner
Workflows 2025-12-05

Z-Image-ControlNet + SDPOSE + second-order Refiner

This workflow uses the enhanced Z-Image Turbo model together with ControlNet to achieve stable control over lines, depth maps, and pose skeletons, covering nearly all common conditioning types. Z-Image Turbo remains extremely fast while respecting the structural guidance extracted from SD-Pose or other detectors, making the first-pass generation both pose-accurate and responsive to the control signal. Because ControlNet naturally lowers aesthetic richness, the workflow adds a secondary refinement stage: the image is decoded, re-encoded into latent space, and passed through an additional diffusion round. This brings back detail, color depth, and overall visual quality without breaking the control. Latent-space upscaling often produces more pleasing variation compared with direct pixel upscaling, so the workflow uses it as the primary refinement strategy. In practice, results show that skeleton control is stable and predictable, and switching to SD-Pose improves the accuracy of complex or full-body detections. For more stylized or line-art inputs, preprocessing strength directly influences how strictly the model interprets the structure, and adjusting it provides smoother control than relying solely on ControlNet’s strength value. ControlNet’s presence can weaken LoRA effects, so users may choose carefully when applying style models. Depth extraction, line control, and even typography guidance work reliably, though aesthetic quality still benefits from the two-stage refinement. Given that the Z-Image ecosystem is new and evolving, this workflow provides a practical balance: strong structural control, fast generation, and a refiner step that restores the vivid aesthetics Z-Image is known for. 🎥 YouTube Video Tutorial Want to know what this workflow actually does and how to start fast? This video explains what the tool is, how to launch the workflow instantly, and shares my core design logic — no local setup, no complicated environment. Everything starts directly on RunningHub, so you can experience it in action first. 👉 YouTube Tutorial： https://youtu.be/7GtFnr8ixnM Before you begin, I recommend watching the video thoroughly — getting the full context helps you understand the tool faster and avoid common detours. ⚙️ RunningHub Workflow Try the workflow online right now — no installation required. 👉 Workflow： https://www.runninghub.ai/post/199659

ZImageTurbo 214 downloads
View public details
sam3 components
Workflows 2025-12-04

sam3 components

This workflow is built around SAM3, providing a flexible set of image and video segmentation tools that can be inserted into any ComfyUI pipeline. SAM3 extends the functionality of earlier segmentation models with a more unified detection system: you can guide segmentation through text prompts, bounding boxes, or point clicks, and the model then extracts the corresponding mask with strong consistency. For videos, SAM3 introduces session-based propagation, allowing an object identified in a single frame to be tracked and masked through the entire sequence. This turns SAM3 into a general-purpose region extraction and tracking module that works equally well for static images and dynamic scenes. The workflow presents all these components in a streamlined layout so the segmentation results can be directly previewed or fed into downstream processing. 🎥 YouTube Video Tutorial Want to know what this workflow actually does and how to start fast? This video explains what the tool is, how to launch the workflow instantly, and shares my core design logic — no local setup, no complicated environment. Everything starts directly on RunningHub, so you can experience it in action first. 👉 YouTube Tutorial： https://youtu.be/Ac0662XPbyE Before you begin, I recommend watching the video thoroughly — getting the full context helps you understand the tool faster and avoid common detours。 ⚙️ RunningHub Workflow Try the workflow online right now — no installation required. 👉 Workflow： https://www.runninghub.ai/post/1993229787548368897/?inviteCode=rh-v1111 If the results meet your expectations, you can later deploy it locally for customization. 🎁 Fan Benefits： Register to get 1000 points + daily login 100 points — enjoy 4090 performance and 48 GB super power! 📺 Bilibili Updates (Mainland China & Asia-Pacific) If you’re in the Asia-Pacific region, you can watch the video below to see the workflow demonstration and creative breakdown. 📺 Bilibili Video： https://www.bilibili.com/video/BV1iZUDBxELi/ ☕ Support Me on Ko-fi If you find my content helpful and want to support future creations, you can buy me a coffee ☕。 Every bit of support helps me keep creating — just like a spark that can ignite a blazing flame。 👉 Ko-fi： https://ko-fi.com/aiksk 💼 Business Contact For collaboration or inquiries, please contact aiksk95 on WeChat。 🎥 YouTube 视频教程 想了解这个工

SD 1.5 317 downloads
View public details
Flux.2 image blending
Workflows 2025-12-04

Flux.2 image blending

This workflow uses Flux 2 as a high-quality blending and image-editing model, combining multiple inputs through controlled composition before passing the merged result into Flux’s diffusion pipeline. The model’s cinematic lighting and strong sense of realism make it well-suited for background replacement, relighting, and structural edits where natural visual integration matters. Quantized variants such as Q6 reduce VRAM use without heavily impacting quality, while SageAttention provides a noticeable speed boost. The compositor prepares the blended canvas, and Flux 2 then interprets the merged scene as a single coherent image, allowing edits that look organically fused instead of artificially pasted. In practice, the workflow works well for realistic edits like relighting, replacing environments, or transforming subjects into stylized versions while keeping overall form intact. Flux 2 handles single-image editing with strong stability, producing consistent results even at high resolutions such as 1536×1536. Its realism gives it an advantage for photographic or cinematic outputs, while the blending stage allows flexible assembly of multiple elements—even if some similarity is naturally lost when using many references. For users who don’t require strict multi-image consistency, it offers a smooth and visually convincing way to combine images and apply Flux 2’s expressive lighting and rendering strengths. 🎥 YouTube Video Tutorial Want to know what this workflow actually does and how to start fast? This video explains what the tool is, how to launch the workflow instantly, and shares my core design logic — no local setup, no complicated environment. Everything starts directly on RunningHub, so you can experience it in action first. 👉 YouTube Tutorial： https://youtu.be/-MWq5srHU0E Before you begin, I recommend watching the video thoroughly — getting the full context helps you understand the tool faster and avoid common detours. ⚙️ RunningHub Workflow Try the workflow online right now — no installation required. 👉 Workflow： https://www.runninghub.ai/post/1995426699946995714/?inviteCode=rh-v1111 If the results meet your expectations, you can later deploy it locally for customization. 🎁 Fan Benefits： Register to get 1000 points + daily login 100 points — enjoy 4090 performance and 48 GB super power! 📺 Bilibili Updates (Mainland China & Asia-Pa

Flux.2 D 147 downloads
View public details
Z-Image-Qwen3vl cleaning & upscaling
Workflows 2025-12-04

Z-Image-Qwen3vl cleaning & upscaling

This workflow pairs the Z-Image model with Qwen3-VL to create a fast system for image washing, variation, and controlled reinterpretation. Z-Image handles the visual transformation, while Qwen3-VL reads the input image and generates descriptive text that feeds back into the workflow. This loop allows the model to produce results that stay close to the original structure but explore new style directions. The turbo version of Z-Image speeds up generation significantly, making rapid testing and creative exploration practical without sacrificing the expressive quality of the output. After generating a variant, the workflow uses a tiled enhancement stage to produce a clean high-resolution result. Splitting the image into several tiles helps preserve texture and avoid the blurring that comes from direct upscaling. A light prompt during this stage prevents over-interpretation and keeps the upscale faithful to the generated image. Together, Qwen3-VL provides intelligent guidance, Z-Image creates fast visual transformations, and the tiled refinement step ensures sharp final output. 🎥 YouTube Video Tutorial Want to know what this workflow actually does and how to start fast? This video explains what the tool is, how to launch the workflow instantly, and shares my core design logic — no local setup, no complicated environment. Everything starts directly on RunningHub, so you can experience it in action first. 👉 YouTube Tutorial： https://youtu.be/Q8hkuyVK0zU Before you begin, I recommend watching the video thoroughly — getting the full context helps you understand the tool faster and avoid common detours. ⚙️ RunningHub Workflow Try the workflow online right now — no installation required. 👉 Workflow： https://www.runninghub.ai/post/1994350779123240961/?inviteCode=rh-v1111 If the results meet your expectations, you can later deploy it locally for customization. 🎁 Fan Benefits： Register to get 1000 points + daily login 100 points — enjoy 4090 performance and 48 GB super power! 📺 Bilibili Updates (Mainland China & Asia-Pacific) If you’re in the Asia-Pacific region, you can watch the video below to see the workflow demonstration and creative breakdown。 📺 Bilibili Video： https://www.bilibili.com/video/BV1KgUfBhEom/ ☕ Support Me on Ko-fi If you find my content helpful and want to support future creations, you can buy me a coffee ☕。 Every bit of

ZImageTurbo 203 downloads
View public details
Wildcards 2025-12-04

SteadyDancer + SDpose: one portrait + one pose (skeleton)

This workflow uses SteadyDancer as a pose-to-video system designed to preserve the identity and proportions of the reference image, even when the supplied skeleton video contains extreme or stylized movements. Unlike earlier pose-driven pipelines where the character often distorted or snapped to the skeleton, SteadyDancer treats the reference as an anchor and generates motion around it, maintaining facial structure and overall silhouette. A Wan2.1-based accelerated i2v model handles the actual video generation, while the first frame is forced to match the reference image to stabilize identity. This makes the method suitable for characters with unusual proportions—such as chibi figures or robots—without needing pose remapping or heavy preprocessing. In practical use, most of the variation comes from the quality and realism of the skeleton video. Stronger pose strength increases motion but can reduce tracking in extreme sideways rotations, though identity remains intact. When paired with SD-Pose, the system demonstrates that tracking issues are not caused by inaccurate skeletons but by the natural limits of extreme motion. For normal human motion, SteadyDancer produces stable, consistent results in both realistic and stylized outputs, generating full-body completions when the reference lacks lower-body information. Prompts still influence the final appearance—especially when generating imaginary elements like robotic limbs—so guiding text remains useful. Overall, the workflow offers a reliable way to animate a single image with good identity retention and smooth motion, even across diverse artistic styles. 🎥 YouTube Video Tutorial Want to know what this workflow actually does and how to start fast? This video explains what the tool is, how to launch the workflow instantly, and shares my core design logic — no local setup, no complicated environment. Everything starts directly on RunningHub, so you can experience it in action first. 👉 YouTube Tutorial： https://youtu.be/7Bk_-gCUVVQ Before you begin, I recommend watching the video thoroughly — getting the full context helps you understand the tool faster and avoid common detours. ⚙️ RunningHub Workflow Try the workflow online right now — no installation required. 👉 Workflow： https://www.runninghub.ai/post/1995841055381762050/?inviteCode=rh-v1111 If the results meet your expectations, you can lat

Wan Video 14B i2v 720p 62 downloads
View public details
Workflows 2025-11-27

FFGO: unlocking the multi-image reference potential of Wan2.2

This workflow uses the lightweight FFGO model as a way to unlock the multi-image reference potential hidden inside Wan2.2. While many reference-based models exist, very few are specifically aligned with the 2.2 architecture, and FFGO doesn’t change the style—it simply makes Wan2.2 more willing to follow multiple reference images simultaneously. By merging several reference elements into a single composite first frame, the model receives a concentrated bundle of structural and visual cues, and Wan2.2 then completes the missing areas based on its own internal capability. The method does not rely on training biases, and it does not enforce a specific direction; if there is any bias at all, it comes from Wan2.2 itself. High-noise and low-noise LoRAs are not recommended, and accelerated LoRAs can noticeably reduce quality, so the workflow remains intentionally restrained to preserve the core strengths of the base model. In practice, the entire pipeline is still just a standard Wan2.2 image-to-video process—the only difference is that the input becomes a fused first frame made from multiple images. This direct fusion avoids the quality loss that would normally come from splitting elements, running them through editing models, and recompressing them before video generation. Wan2.2 performs best at a recommended resolution of 1280×720 (or the reverse) and around 81 frames, producing the clearest and most stable results. The first few frames of the output typically reflect the reference composition rather than the actual motion, so trimming out the beginning is expected. Compared with older Wan2.1 reference methods such as Phantom, MagRef, or BindWeave—which often produced limited motion or reduced similarity—this approach is far more straightforward and reliable. By combining a clean composite first frame with FFGO’s ability to trigger Wan2.2’s internal reference behavior, the workflow offers a practical and effective way to generate videos that keep multiple elements integrated with high similarity and minimal quality loss. 🎥 YouTube Video Tutorial Want to know what this workflow actually does and how to start fast? This video explains what the tool is, how to launch the workflow instantly, and shares my core design logic — no local setup, no complicated environment. Everything starts directly on RunningHub, so you can experience it in action first. 👉 YouTube

Wan Video 2.2 I2V-A14B 562 downloads
View public details
Workflows 2025-11-27

HunyuanVideo 1.5 accelerated image-to-video 1080p

This workflow focuses on accelerating image-to-video generation with HunyuanVideo 1.5 by combining its native i2v diffusion model with a four-step Light-X2V acceleration LoRA. The accelerated LoRA compresses the long inference path of the standard model into a much shorter sequence, so a 720p clip that normally requires more than 30 minutes can be produced in about 10 minutes. Even though Light-X2V is labeled as a text-to-video accelerator, it still works reliably for image-to-video when paired with Hunyuan’s architecture. The workflow keeps the structure clean: an uploaded start image feeds into Hunyuan’s dual-encoder system, then the accelerated sampling produces the motion sequence, and the final 1080p enhancement remains available as an optional finishing stage. Despite Hunyuan being less universally adaptable than Wan2.2, its training quality is still excellent, making it well-suited for fast, high-resolution image-to-video tasks. In use, the workflow is straightforward—upload an image, write the prompt, set resolution and frame length, and run the accelerated pipeline. Lower CFG values help stabilize the accelerated model, and although it can run at four steps, many users prefer keeping it around eight for more dependable motion. Reducing LoRA strength slightly prevents over-pushing the dynamics, especially when using the 2-rank or 8-rank variants. The Sage Attention module can speed up model interaction without changing the visual style, and the final tiled 1080p decode stage preserves clarity while keeping the upscale process automatic. Motion intensity may drop slightly at extreme acceleration, but overall the speed increase is substantial, and the quality remains steady enough for practical production. For creators who need quick, clean 720p–1080p video from a single image, this accelerated setup offers an efficient, reliable workflow built on top of HunyuanVideo 1.5’s strengths. 🎥 YouTube Video Tutorial Want to know what this workflow actually does and how to start fast? This video explains what the tool is, how to launch the workflow instantly, and shares my core design logic — no local setup, no complicated environment. Everything starts directly on RunningHub, so you can experience it in action first. 👉 YouTube Tutorial： https://youtu.be/Ac0662XPbyE Before you begin, I recommend watching the video thoroughly — getting the full context h

Hunyuan Video 255 downloads
View public details
Magic-Wan-V2 accelerated tiled ultra-resolution upscaling
Workflows 2025-11-23

Magic-Wan-V2 accelerated tiled ultra-resolution upscaling

This workflow is built to combine the speed of the accelerated Wan 2.2 four-step model with the precision of multi-tile ultra-resolution reconstruction. The image is first processed through Magic Wan 2.0’s writing-realism model, then passed into a tiling pipeline where the picture is cut into sections, upscaled, and decoded tile-by-tile before being reassembled into a single high-resolution image. This approach avoids the usual blurring and texture collapse that happen with direct large-scale upscaling, and instead allows each tile to be optimized locally with much cleaner detail. The result is a fast yet high-clarity workflow that can push images to significantly higher resolutions while still keeping the core structure and style intact. When using this setup, the key is simply choosing a reasonable tile layout—splits like 2×2 or 3×3 typically provide the best balance between clarity and stability. Because each tile is processed independently, giving the model a clear visual direction through the prompt helps maintain consistency across the different regions. The separate pixel-based upscale step also contributes to smoother transitions before tiling begins, and avoids pushing tiles too far outside the model’s comfortable range. Overall, this workflow offers a reliable way to get very large, very clean results quickly, blending accelerated generation with precise tiled reconstruction for high-detail final output. 🎥 YouTube Video Tutorial Want to know what this workflow actually does and how to start fast? This video explains what the tool is, how to launch the workflow instantly, and shares my core design logic — no local setup, no complicated environment. Everything starts directly on RunningHub, so you can experience it in action first. 👉 YouTube Tutorial： https://youtu.be/Eek663BSp3g Before you begin, I recommend watching the video thoroughly — getting the full context helps you understand the tool faster and avoid common detours。 ⚙️ RunningHub Workflow Try the workflow online right now — no installation required. 👉 Workflow： https://www.runninghub.ai/post/1991145078443868162/?inviteCode=rh-v1111 If the results meet your expectations, you can later deploy it locally for customization. 🎁 Fan Benefits： Register to get 1000 points + daily login 100 points — enjoy 4090 performance and 48 GB super power! 📺 Bilibili Updates (Mainland Ch

Wan Video 2.2 T2V-A14B 203 downloads
View public details
Magic-Wan-V2 image-to-image
Workflows 2025-11-23

Magic-Wan-V2 image-to-image

This workflow is built around Magic Wan 2.0, a realism-oriented model that handles detail refinement, texture consistency, and lighting reconstruction particularly well when working from an input image. The structure here takes an uploaded image, resizes it to a clean working resolution, encodes it into latent space through the Wan VAE, and then lets the model re-interpret that content under the guidance of your prompt. Everything is arranged so that you can drop in an image, add the direction you want, and immediately generate a refined or re-styled result without having to manage the more technical components of the pipeline. When using it, a few practical points help the output stay stable and high-quality. Keeping moderate to slightly higher steps maintains good fidelity without drifting away from the original input. Denoise levels are especially important in image-to-image: lower values help preserve structure, while higher ones push stronger changes—so treating denoise as the “strength slider” gives you more control. If you’re planning to upscale afterwards, the same low-denoise approach keeps consistency across details. And as with the text-to-image workflow, acceleration LoRAs aren’t recommended since they noticeably weaken realism. Overall, it’s a straightforward process: adjust denoise based on how much transformation you want, set your prompt direction, and the workflow handles the rest for a balanced, high-quality reconstruction. 🎥 YouTube Video Tutorial Want to know what this workflow actually does and how to start fast? This video explains what the tool is, how to launch the workflow instantly, and shares my core design logic — no local setup, no complicated environment. Everything starts directly on RunningHub, so you can experience it in action first. 👉 YouTube Tutorial： https://youtu.be/Eek663BSp3g Before you begin, I recommend watching the video thoroughly — getting the full context helps you understand the tool faster and avoid common detours. ⚙️ RunningHub Workflow Try the workflow online right now — no installation required. 👉 Workflow： https://www.runninghub.ai/post/1991113183234732033/?inviteCode=rh-v1111 If the results meet your expectations, you can later deploy it locally for customization. 🎁 Fan Benefits： Register to get 1000 points + daily login 100 points — enjoy 4090 performance and 48 GB super power! 📺 B

Wan Video 2.2 T2V-A14B 208 downloads
View public details
Magic-Wan-V2 accelerated image-to-image
Workflows 2025-11-23

Magic-Wan-V2 accelerated image-to-image

This workflow is designed to let Magic Wan 2.0 handle image-to-image tasks at high speed by combining it with a four-step acceleration model derived from Wan 2.2. The input image is first resized and encoded into the latent space, and then the accelerated model performs reconstruction with minimal sampling steps. This makes the process extremely fast while still allowing the prompt to influence the atmosphere, lighting, and overall direction of the rewritten image. The workflow keeps the structure of a standard img2img pipeline, but the accelerated backbone gives it a much quicker turnaround, making it suitable for rapid style adjustments, fast reinterpretations, and quick visual experiments based on an existing image. To get the best results with this approach, clear prompt direction helps the model maintain coherence, since the compressed sampling path relies more heavily on textual guidance. Adding explicit cues about style, mood, lighting, or material texture can significantly improve the stability of the output. Because the model is optimized for speed, the reconstruction focuses on broad visual consistency rather than micro-level detail, so this workflow is ideal for quickly exploring variations of a source image. Once you find a direction you like, you can move to a slower high-quality workflow for refinement, while this accelerated version remains a fast and efficient way to iterate through creative options. 🎥 YouTube Video Tutorial Want to know what this workflow actually does and how to start fast? This video explains what the tool is, how to launch the workflow instantly, and shares my core design logic — no local setup, no complicated environment. Everything starts directly on RunningHub, so you can experience it in action first. 👉 YouTube Tutorial： https://youtu.be/Eek663BSp3g Before you begin, I recommend watching the video thoroughly — getting the full context helps you understand the tool faster and avoid common detours。 ⚙️ RunningHub Workflow Try the workflow online right now — no installation required. 👉 Workflow： https://www.runninghub.ai/post/1991143691563696129/?inviteCode=rh-v1111 If the results meet your expectations, you can later deploy it locally for customization. 🎁 Fan Benefits： Register to get 1000 points + daily login 100 points — enjoy 4090 performance and 48 GB super power! 📺 Bilibili Updates (Mainland China

Wan Video 2.2 T2V-A14B 133 downloads
View public details
Magic-Wan-V2 accelerated text-to-image
Workflows 2025-11-23

Magic-Wan-V2 accelerated text-to-image

This workflow combines the realistic rendering capability of Magic Wan 2.0 with a four-step acceleration model derived from Wan 2.2, allowing text-to-image generation to run at significantly higher speed. The accelerated model compresses the usual multi-step sampling process into a short convergence path, giving you near-instant results while preserving the core characteristics of the Magic Wan aesthetic. The pipeline itself remains the same—prompt encoding, latent generation, and VAE decoding—so the workflow still behaves like a standard text-to-image process, just much faster. It’s most suitable for rapid ideation, quick composition testing, and fast turnaround scenarios where you need usable results without waiting for long inference times. When working with this accelerated setup, providing clear and direct prompts helps the model maintain structure and realism despite the reduced sampling steps. Because the acceleration model emphasizes speed over micro-detail, adding precise cues about lighting, material textures, scene layout, or subject posture can noticeably improve stability. The step count is already minimized, and keeping it at or slightly above the default ensures a balance between responsiveness and visual consistency. Overall, this workflow functions as a fast exploratory mode—ideal for quickly iterating through concepts—after which you can switch to a full-quality, non-accelerated pipeline for final rendering if needed. 🎥 YouTube Video Tutorial Want to know what this workflow actually does and how to start fast? This video explains what the tool is, how to launch the workflow instantly, and shares my core design logic — no local setup, no complicated environment. Everything starts directly on RunningHub, so you can experience it in action first. 👉 YouTube Tutorial： https://youtu.be/Eek663BSp3g Before you begin, I recommend watching the video thoroughly — getting the full context helps you understand the tool faster and avoid common detours。 ⚙️ RunningHub Workflow Try the workflow online right now — no installation required. 👉 Workflow： https://www.runninghub.ai/post/1991141173844639745/?inviteCode=rh-v1111 If the results meet your expectations, you can later deploy it locally for customization. 🎁 Fan Benefits： Register to get 1000 points + daily login 100 points — enjoy 4090 performance and 48 GB super power! 📺 Bilibili Upd

Wan Video 2.2 T2V-A14B 112 downloads
View public details
Magic-Wan-V2 text-to-image
Workflows 2025-11-23

Magic-Wan-V2 text-to-image

Magic Wan 2.0 is a realism-oriented generation model that focuses on natural skin texture, lighting accuracy, and fine surface details. Its output tends to be clean, stable, and close to real photographic aesthetics. The workflow I assembled is a fully ready-to-use text-to-image chain, so the prompt goes in and the system handles the encoding and generation without requiring you to build or rearrange nodes. It’s designed to keep the pipeline consistent and reliable, making it suitable for portraits, real-world scenes, and any kind of detail-heavy realistic imagery. For practical use, there are a few simple but very effective tips. Using slightly higher steps gives noticeably better stability and richer detail—around the low-20s is a good balance unless you’re intentionally optimizing for speed. This model also benefits a lot from tile-based upscaling; a 2×2 split when generating 4K results keeps the texture much cleaner and avoids smearing. The denoise value shouldn’t be set too high, especially during upscaling—around 0.2 to 0.3 is already very reactive and usually enough. And even though turbo or “4-step” acceleration LoRAs can speed up generation, they significantly damage realism, so they’re not recommended for this workflow. Overall, it’s a straightforward setup that still delivers high-quality, realistic images without much effort. 🎥 YouTube Video Tutorial Want to know what this workflow actually does and how to start fast? This video explains what the tool is, how to launch the workflow instantly, and shares my core design logic — no local setup, no complicated environment. Everything starts directly on RunningHub, so you can experience it in action first. 👉 YouTube Tutorial： https://youtu.be/Eek663BSp3g Before you begin, I recommend watching the video thoroughly — getting the full context helps you understand the tool faster and avoid common detours. ⚙️ RunningHub Workflow Try the workflow online right now — no installation required. 👉 Workflow： https://www.runninghub.ai/post/1990870633426919425/?inviteCode=rh-v1111 If the results meet your expectations, you can later deploy it locally for customization. 🎁 Fan Benefits： Register to get 1000 points + daily login 100 points — enjoy 4090 performance and 48 GB super power! 📺 Bilibili Updates (Mainland China & Asia-Pacific) If you’re in the Asia-Pacific region, you can watch the video

Wan Video 2.2 T2V-A14B 153 downloads
View public details
Magic-Wan-V2 tiled ultra-resolution upscaling
Workflows 2025-11-19

Magic-Wan-V2 tiled ultra-resolution upscaling

This workflow is designed specifically to handle ultra-resolution upscaling using a tiled reconstruction method. Instead of enlarging the entire image at once—which often leads to smearing, blurred textures, or broken details—it splits the picture into multiple tiles, enhances each tile independently, and then reassembles them with precise positional data. Magic Wan 2.0 provides the underlying reconstruction ability, while the tile cutting, tile decoding, and tile assembly steps ensure that each section is processed with higher local clarity. The result is an image that keeps the original structure intact but gains significantly sharper textures, more stable edges, and overall a much cleaner high-resolution appearance. Using the workflow effectively mainly comes down to a few intuitive points. Choosing a reasonable tile count—like a 2×2 or 3×3 split—usually delivers the clearest result without causing visible seams. Lower denoise values help maintain detail consistency between tiles and prevent the model from over-rewriting local regions, especially at very large output sizes. Pre-scaling the image slightly before tiling can improve texture richness, but overscaling will only introduce unnecessary variability. As long as the tiles stay within a size the model handles well, the workflow will take care of the rest, producing an image that is noticeably sharper and more detailed while still preserving the original content and lighting. 🎥 YouTube Video Tutorial Want to know what this workflow actually does and how to start fast? This video explains what the tool is, how to launch the workflow instantly, and shares my core design logic — no local setup, no complicated environment. Everything starts directly on RunningHub, so you can experience it in action first. 👉 YouTube Tutorial： https://youtu.be/Eek663BSp3g Before you begin, I recommend watching the video thoroughly — getting the full context helps you understand the tool faster and avoid common detours. ⚙️ RunningHub Workflow Try the workflow online right now — no installation required. 👉 Workflow： https://www.runninghub.ai/post/1991113618789892098/?inviteCode=rh-v1111 If the results meet your expectations, you can later deploy it locally for customization. 🎁 Fan Benefits： Register to get 1000 points + daily login 100 points — enjoy 4090 performance and 48 GB super power! 📺 Bilibili U

Wan Video 2.2 T2V-A14B 207 downloads
View public details
Workflows 2025-11-18

MoCha: character replacement like Viggle

🎥 YouTube Video Tutorial Want to know what this workflow actually does and how to start fast? This video explains what the tool is, how to launch the workflow instantly, and shares my core design logic — no local setup, no complicated environment. Everything starts directly on RunningHub, so you can experience it in action first. 👉 YouTube Tutorial： https://youtu.be/OYnCPNDX-1Q Before you begin, I recommend watching the video thoroughly — getting the full context helps you understand the tool faster and avoid common detours. ⚙️ RunningHub Workflow Try the workflow online right now — no installation required. 👉 Workflow： https://www.runninghub.ai/post/1981408974144708609/?inviteCode=rh-v1111 If the results meet your expectations, you can later deploy it locally for customization. 🎁 Fan Benefits： Register to get 1000 points + daily login 100 points — enjoy 4090 performance and 48 GB super power! 📺 Bilibili Updates (Mainland China & Asia-Pacific) If you’re in the Asia-Pacific region, you can watch the video below to see the workflow demonstration and creative breakdown. 📺 Bilibili Video： https://www.bilibili.com/video/BV1nmsnzMEJi/ ☕ Support Me on Ko-fi If you find my content helpful and want to support future creations, you can buy me a coffee ☕. Every bit of support helps me keep creating — just like a spark that can ignite a blazing flame. 👉 Ko-fi： https://ko-fi.com/aiksk 💼 Business Contact For collaboration or inquiries, please contact aiksk95 on WeChat. 🎥 YouTube 视频教程 想了解这个工作流到底是怎样的工具，以及如何快速启动？ 视频主要介绍 工具定位、快速启动方法 和 我的构筑思路 。 我们会直接在 RunningHub 上进行演示，让你第一时间看到实际效果。 👉 YouTube 教程： https://youtu.be/OYnCPNDX-1Q 开始前建议尽量完整地观看视频 —— 把握整体思路会更快上手，也能少走常见弯路。 ⚙️ 在线体验工作流 现在就可以在线体验，无需安装。 👉 工作流： https://www.runninghub.ai/post/1981408974144708609/?inviteCode=rh-v1111 打开上方链接即可直接运行该工作流，实时查看生成效果。 如果觉得效果理想，你也可以在本地进行自定义部署。 🎁 粉丝福利： 注册即送 1000 积分 ，每日登录 100 积分 ，畅玩 4090 体验 48 G 超级性能！ 📺 Bilibili 更新（中国大陆及南亚太地区） 如果你在中国大陆或南亚太地区，可以通过下方视频查看该工作流的实测效果与构思讲解。 📺 B站视频： https://www.bilibili.com/video/BV1nmsnzMEJi/ 我会在 夸克网盘 持续更新模型资源： 👉 https://pan.quark.cn/s/20c6f6f8d87b 这些资源主要面向本地用户，方便进行创作与学习。

Wan Video 14B t2v 238 downloads
View public details
Workflows 2025-11-18

Wan2.2 image-to-video ultra cost-effective acceleration V4

🎥 YouTube Video Tutorial Want to know what this workflow actually does and how to start fast? This video explains what the tool is, how to launch the workflow instantly, and shares my core design logic — no local setup, no complicated environment. Everything starts directly on RunningHub, so you can experience it in action first. 👉 YouTube Tutorial： https://youtu.be/P62bHKK1Pa0 Before you begin, I recommend watching the video thoroughly — getting the full context helps you understand the tool faster and avoid common detours. ⚙️ RunningHub Workflow Try the workflow online right now — no installation required. 👉 Workflow： https://www.runninghub.ai/post/1978022050193362945/?inviteCode=rh-v1111 If the results meet your expectations, you can later deploy it locally for customization. 🎁 Fan Benefits： Register to get 1000 points + daily login 100 points — enjoy 4090 performance and 48 GB super power! 📺 Bilibili Updates (Mainland China & Asia-Pacific) If you’re in the Asia-Pacific region, you can watch the video below to see the workflow demonstration and creative breakdown. 📺 Bilibili Video： https://www.bilibili.com/video/BV1UF4tz3EuE/ ☕ Support Me on Ko-fi If you find my content helpful and want to support future creations, you can buy me a coffee ☕. Every bit of support helps me keep creating — just like a spark that can ignite a blazing flame. 👉 Ko-fi： https://ko-fi.com/aiksk 💼 Business Contact For collaboration or inquiries, please contact aiksk95 on WeChat. 🎥 YouTube 视频教程 想了解这个工作流到底是怎样的工具，以及如何快速启动？ 视频主要介绍 工具定位、快速启动方法 和 我的构筑思路 。 我们会直接在 RunningHub 上进行演示，让你第一时间看到实际效果。 👉 YouTube 教程： https://youtu.be/P62bHKK1Pa0 开始前建议尽量完整地观看视频 —— 把握整体思路会更快上手，也能少走常见弯路。 ⚙️ 在线体验工作流 现在就可以在线体验，无需安装。 👉 工作流： https://www.runninghub.ai/post/1978022050193362945/?inviteCode=rh-v1111 打开上方链接即可直接运行该工作流，实时查看生成效果。 如果觉得效果理想，你也可以在本地进行自定义部署。 🎁 粉丝福利： 注册即送 1000 积分 ，每日登录 100 积分 ，畅玩 4090 体验 48 G 超级性能！ 📺 Bilibili 更新（中国大陆及南亚太地区） 如果你在中国大陆或南亚太地区，可以通过下方视频查看该工作流的实测效果与构思讲解。 📺 B站视频： https://www.bilibili.com/video/BV1UF4tz3EuE/ 我会在 夸克网盘 持续更新模型资源： 👉 https://pan.quark.cn/s/20c6f6f8d87b 这些资源主要面向本地用户，方便进行创作与学习。

Wan Video 2.2 I2V-A14B 152 downloads
View public details
Qwen-Image-Edit-Rapid-AIO image editing
Workflows 2025-11-18

Qwen-Image-Edit-Rapid-AIO image editing

🎥 YouTube Video Tutorial Want to know what this workflow actually does and how to start fast? This video explains what the tool is, how to launch the workflow instantly, and shares my core design logic — no local setup, no complicated environment. Everything starts directly on RunningHub, so you can experience it in action first. 👉 YouTube Tutorial： https://youtu.be/nP_tINJ0N0A Before you begin, I recommend watching the video thoroughly — getting the full context helps you understand the tool faster and avoid common detours. ⚙️ RunningHub Workflow Try the workflow online right now — no installation required. 👉 Workflow： https://www.runninghub.ai/post/1977650991560577025/?inviteCode=rh-v1111 If the results meet your expectations, you can later deploy it locally for customization. 🎁 Fan Benefits： Register to get 1000 points + daily login 100 points — enjoy 4090 performance and 48 GB super power! 📺 Bilibili Updates (Mainland China & Asia-Pacific) If you’re in the Asia-Pacific region, you can watch the video below to see the workflow demonstration and creative breakdown. 📺 Bilibili Video： https://www.bilibili.com/video/BV1wN42zDEDK/ ☕ Support Me on Ko-fi If you find my content helpful and want to support future creations, you can buy me a coffee ☕. Every bit of support helps me keep creating — just like a spark that can ignite a blazing flame. 👉 Ko-fi： https://ko-fi.com/aiksk 💼 Business Contact For collaboration or inquiries, please contact aiksk95 on WeChat. 🎥 YouTube 视频教程 想了解这个工作流到底是怎样的工具，以及如何快速启动？ 视频主要介绍 工具定位、快速启动方法 和 我的构筑思路 。 我们会直接在 RunningHub 上进行演示，让你第一时间看到实际效果。 👉 YouTube 教程： https://youtu.be/nP_tINJ0N0A 开始前建议尽量完整地观看视频 —— 把握整体思路会更快上手，也能少走常见弯路。 ⚙️ 在线体验工作流 现在就可以在线体验，无需安装。 👉 工作流： https://www.runninghub.ai/post/1977650991560577025/?inviteCode=rh-v1111 打开上方链接即可直接运行该工作流，实时查看生成效果。 如果觉得效果理想，你也可以在本地进行自定义部署。 🎁 粉丝福利： 注册即送 1000 积分 ，每日登录 100 积分 ，畅玩 4090 体验 48 G 超级性能！ 📺 Bilibili 更新（中国大陆及南亚太地区） 如果你在中国大陆或南亚太地区，可以通过下方视频查看该工作流的实测效果与构思讲解。 📺 B站视频： https://www.bilibili.com/video/BV1wN42zDEDK/ 我会在 夸克网盘 持续更新模型资源： 👉 https://pan.quark.cn/s/20c6f6f8d87b 这些资源主要面向本地用户，方便进行创作与学习。

Qwen 816 downloads
View public details
Workflows 2025-11-18

Flash VSR video super-resolution

🎥 YouTube Video Tutorial Want to know what this workflow actually does and how to start fast? This video explains what the tool is, how to launch the workflow instantly, and shares my core design logic — no local setup, no complicated environment. Everything starts directly on RunningHub, so you can experience it in action first. 👉 YouTube Tutorial： https://youtu.be/4MXsqE1uYL4 Before you begin, I recommend watching the video thoroughly — getting the full context helps you understand the tool faster and avoid common detours. ⚙️ RunningHub Workflow Try the workflow online right now — no installation required. 👉 Workflow： https://www.runninghub.ai/post/1979451793736372225/?inviteCode=rh-v1111 If the results meet your expectations, you can later deploy it locally for customization. 🎁 Fan Benefits： Register to get 1000 points + daily login 100 points — enjoy 4090 performance and 48 GB super power! 📺 Bilibili Updates (Mainland China & Asia-Pacific) If you’re in the Asia-Pacific region, you can watch the video below to see the workflow demonstration and creative breakdown. 📺 Bilibili Video： https://www.bilibili.com/video/BV1RoWxzkEoc/ ☕ Support Me on Ko-fi If you find my content helpful and want to support future creations, you can buy me a coffee ☕. Every bit of support helps me keep creating — just like a spark that can ignite a blazing flame. 👉 Ko-fi： https://ko-fi.com/aiksk 💼 Business Contact For collaboration or inquiries, please contact aiksk95 on WeChat. 🎥 YouTube 视频教程 想了解这个工作流到底是怎样的工具，以及如何快速启动？ 视频主要介绍 工具定位、快速启动方法 和 我的构筑思路 。 我们会直接在 RunningHub 上进行演示，让你第一时间看到实际效果。 👉 YouTube 教程： https://youtu.be/4MXsqE1uYL4 开始前建议尽量完整地观看视频 —— 把握整体思路会更快上手，也能少走常见弯路。 ⚙️ 在线体验工作流 现在就可以在线体验，无需安装。 👉 工作流： https://www.runninghub.ai/post/1979451793736372225/?inviteCode=rh-v1111 打开上方链接即可直接运行该工作流，实时查看生成效果。 如果觉得效果理想，你也可以在本地进行自定义部署。 🎁 粉丝福利： 注册即送 1000 积分 ，每日登录 100 积分 ，畅玩 4090 体验 48 G 超级性能！ 📺 Bilibili 更新（中国大陆及南亚太地区） 如果你在中国大陆或南亚太地区，可以通过下方视频查看该工作流的实测效果与构思讲解。 📺 B站视频： https://www.bilibili.com/video/BV1RoWxzkEoc/ 我会在 夸克网盘 持续更新模型资源： 👉 https://pan.quark.cn/s/20c6f6f8d87b 这些资源主要面向本地用户，方便进行创作与学习。

Wan Video 1.3B t2v 248 downloads
View public details
Qwen-Image-Edit camera control
Workflows 2025-11-18

Qwen-Image-Edit camera control

🎥 YouTube Video Tutorial Want to know what this workflow actually does and how to start fast? This video explains what the tool is, how to launch the workflow instantly, and shares my core design logic — no local setup, no complicated environment. Everything starts directly on RunningHub, so you can experience it in action first. 👉 YouTube Tutorial： https://youtu.be/HJyEOoihUNc Before you begin, I recommend watching the video thoroughly — getting the full context helps you understand the tool faster and avoid common detours. ⚙️ RunningHub Workflow Try the workflow online right now — no installation required. 👉 Workflow： https://www.runninghub.ai/post/1990374981525803009/?inviteCode=rh-v1111 If the results meet your expectations, you can later deploy it locally for customization. 🎁 Fan Benefits： Register to get 1000 points + daily login 100 points — enjoy 4090 performance and 48 GB super power! 📺 Bilibili Updates (Mainland China & Asia-Pacific) If you’re in the Asia-Pacific region, you can watch the video below to see the workflow demonstration and creative breakdown. 📺 Bilibili Video： https://www.bilibili.com/video/BV1HcyGBjEep/ ☕ Support Me on Ko-fi If you find my content helpful and want to support future creations, you can buy me a coffee ☕. Every bit of support helps me keep creating — just like a spark that can ignite a blazing flame. 👉 Ko-fi： https://ko-fi.com/aiksk 💼 Business Contact For collaboration or inquiries, please contact aiksk95 on WeChat. 🎥 YouTube 视频教程 想了解这个工作流到底是怎样的工具，以及如何快速启动？ 视频主要介绍 工具定位、快速启动方法 和 我的构筑思路 。 我们会直接在 RunningHub 上进行演示，让你第一时间看到实际效果。 👉 YouTube 教程： https://youtu.be/HJyEOoihUNc 开始前建议尽量完整地观看视频 —— 把握整体思路会更快上手，也能少走常见弯路。 ⚙️ 在线体验工作流 现在就可以在线体验，无需安装。 👉 工作流： https://www.runninghub.ai/post/1990374981525803009/?inviteCode=rh-v1111 打开上方链接即可直接运行该工作流，实时查看生成效果。 如果觉得效果理想，你也可以在本地进行自定义部署。 🎁 粉丝福利： 注册即送 1000 积分 ，每日登录 100 积分 ，畅玩 4090 体验 48 G 超级性能！ 📺 Bilibili 更新（中国大陆及南亚太地区） 如果你在中国大陆或南亚太地区，可以通过下方视频查看该工作流的实测效果与构思讲解。 📺 B站视频： https://www.bilibili.com/video/BV1HcyGBjEep/ 我会在 夸克网盘 持续更新模型资源： 👉 https://pan.quark.cn/s/20c6f6f8d87b 这些资源主要面向本地用户，方便进行创作与学习。

Qwen 292 downloads
View public details