DiffSynth-Studio
DiffSynth-Studio copied to clipboard
modelscope
Enjoy the magic of Diffusion models!
您好，感谢您的工作，我有关于模型结构修改的两个问题： 1. 在原始Wan2.1-Fun-V1.1-1.3B-Control模型中，模型输入为reference_image（参考图像）, control_video（控制视频），video（视频真值），现在我想将第一帧图像到视频生成模型和原模型结合起来，那么我在extra_inputs中还多添加了一个input_images，这样子在pipeline中会自动调用WanVideoUnit_ImageEmbedderVAE以更新y且不会改变模型结构，这样子是否能够训练一个新的模型 2. 我对wan_video_dit模型作出结构上的修改，比如多增加了一个动作输入，是否在self.keys_hash_with_shape_dict中添加新模型检查点的hashkey就可以完成新模型的注册和使用
before train  after train 
Hello, thank you for your work. May I ask, if I want to add a module to WanVideoUnit, such as adding an action embedding module and combining the action embedding...
**Version: 1.1.8** I am encountering an issue where a LoRA trained on the **Wan2.1-I2V-14B-480P (Image-to-Video)** base model does not produce the expected video output. ### Successful Scenario (Working) * A...
我看到尽管```QwenImagePipeline.__call__```有把tiled, tile_size, tile_stride参数传入self.vae.decode，但```QwenImageVAE.decode```只是用**kwargs接收，并没有实际用到。 https://github.com/modelscope/DiffSynth-Studio/blob/afd101f3452c9ecae0c87b79adfa2e22d65ffdc3/diffsynth/pipelines/qwen_image.py#L444 https://github.com/modelscope/DiffSynth-Studio/blob/afd101f3452c9ecae0c87b79adfa2e22d65ffdc3/diffsynth/models/qwen_image_vae.py#L716 我对模型原理了解有限，想请教一下：请问是 功能暂未实现 还是 QwenImageVAE模型架构本身就无法支持tiled decoding？感谢您的时间和贡献！
Hi, Do you know how to load Qwen-Image-Lightning's acceleration lora in qwen-image-eidt-2509? In DiffSynth-Studio I loaded the acceleration lora and it shows 720 tensors are uploaded by LoRA, but the...
How much data and how many training steps are needed for qwen-image-edit lora?
Hi, I used lora to finetune the Wan2.1 VACE1.3B model with the script `examples/wanvideo/model_training/lora/Wan2.1-VACE-1.3B.sh`. I used only one data pair and trained with all default parameters in the training script....
