是否能夠使用SDXL模型?

[Cooper0607]

使用SDXL的模型 AnimateDiff選擇mm_sdxl_v10_beta.ckpt 會產生錯誤

請問錯誤原因為何??

錯誤代碼如下 謝謝

RuntimeError: Input type (struct c10::Half) and bias type (float) should be the same Traceback: File "D:\DiffSynth-Studio.glut\lib\site-packages\streamlit\runtime\scriptrunner\script_runner.py", line 542, in _run_script exec(code, module.dict) File "D:\DiffSynth-Studio\pages\2_Video_Creator.py", line 197, in SDVideoPipelineRunner(in_streamlit=True).run(config) File "D:\DiffSynth-Studio\diffsynth\pipelines\stable_diffusion_video.py", line 349, in run output_video = self.synthesize_video(model_manager, pipe, config["pipeline"]["seed"], smoother, config["pipeline"]["pipeline_inputs"]) File "D:\DiffSynth-Studio\diffsynth\pipelines\stable_diffusion_video.py", line 296, in synthesize_video output_video = pipe(pipeline_inputs, smoother=smoother, progress_bar_st=progress_bar_st) File "D:\DiffSynth-Studio.glut\lib\site-packages\torch\utils_contextlib.py", line 115, in decorate_context return func(*args, kwargs) File "D:\DiffSynth-Studio\diffsynth\pipelines\stable_diffusion_video.py", line 189, in call latents = self.encode_images(input_frames) File "D:\DiffSynth-Studio\diffsynth\pipelines\stable_diffusion_video.py", line 147, in encode_images latent = self.vae_encoder(image, tiled=tiled, tile_size=tile_size, tile_stride=tile_stride).cpu() File "D:\DiffSynth-Studio.glut\lib\site-packages\torch\nn\modules\module.py", line 1518, in _wrapped_call_impl return self._call_impl(*args, *kwargs) File "D:\DiffSynth-Studio.glut\lib\site-packages\torch\nn\modules\module.py", line 1527, in _call_impl return forward_call(args, kwargs) File "D:\DiffSynth-Studio\diffsynth\models\sd_vae_encoder.py", line 58, in forward hidden_states = self.conv_in(sample) File "D:\DiffSynth-Studio.glut\lib\site-packages\torch\nn\modules\module.py", line 1518, in _wrapped_call_impl return self._call_impl(*args, *kwargs) File "D:\DiffSynth-Studio.glut\lib\site-packages\torch\nn\modules\module.py", line 1527, in _call_impl return forward_call(args, **kwargs) File "D:\DiffSynth-Studio.glut\lib\site-packages\torch\nn\modules\conv.py", line 460, in forward return self._conv_forward(input, self.weight, self.bias) File "D:\DiffSynth-Studio.glut\lib\site-packages\torch\nn\modules\conv.py", line 456, in _conv_forward return F.conv2d(input, weight, bias, self.stride,

[Wenxianxin]

我也遇到过，好像官方还不支持SDXL模型，试了好多LCM的模型都不行，自己又不会写

[Artiprocher]

SDXL 支持 AnimateDiff，但我们不建议用，因为效果实在是太差，我们仅能做到能跑。

https://github.com/modelscope/DiffSynth-Studio/blob/main/examples/video_synthesis/sdxl_text_to_video.py

[Cooper0607]

感謝回復