Skip to content

Commit 210b9ba

Browse files
author
ai-business-hql
committed
feat: expert improve
1 parent 26aa4ed commit 210b9ba

File tree

2 files changed

+3
-2
lines changed

2 files changed

+3
-2
lines changed

‎README_CN.md‎

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -71,6 +71,7 @@ https://github.com/user-attachments/assets/0372faf4-eb64-4aad-82e6-5fd69f349c2c
7171

7272
- 3. 💎 **之前的工作流生图效果不满意？**：提出您不满意的地方，让我们帮您修改工作流，增加节点，修改参数，优化工作流结构。
7373
- 在输入框输入：帮我在当前的画布上添加一个xxxx。
74+
- 注意事项：如果是2025年5月往后新出的模型，例如wan2.2等模型，可能会导致LLM无法理解，进程中断。
7475
<img src="assets/改写.gif"/>
7576

7677
- 4. 💎 **调参过程太痛苦？**：我们为您提供了调参工具，您可以设置参数范围，系统会自动批量执行不同参数组合，并生成结果可视化对比，帮助您快速找到最优参数配置。

‎backend/data/workflow_rewrite_expert.json‎

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -46,7 +46,7 @@
4646
},
4747
{
4848
"name": "wan2.2图生视频",
49-
"description": "wan2.2是图像生成视频的模型,是wan2.1的升级版，主要用来帮助用户使用图片生成视频和根据图片加提示词来进行视频素材的创作,用在comfyui中图片转视频的后期创作。",
50-
"content": "wan2.2将图片转为视频(需要两个wanvideo model loader节点,在这个节点中分别需要高低噪两个视频模型,其模型为HIGH_fp8、LOW_fp8并且需要两个WanVideo Sampler节点来进行分布采样。wan2.2将图片转为视频(需要两个wanvideo model loader节点，在这个节点中分别需要高低噪两个视频模型,其模型为HIGH_fp8、LOW_fp8。vae需要wanvideo vae loader和wan2_1_vae_bf16模型),并需要用到专属的wan系列生图节点组。 {\n \"11\": {\n \"inputs\": {\n \"model_name\": \"umt5-xxl-enc-bf16.safetensors\",\n \"precision\": \"bf16\",\n \"load_device\": \"offload_device\",\n \"quantization\": \"disabled\"\n },\n \"class_type\": \"LoadWanVideoT5TextEncoder\",\n \"_meta\": {\n \"title\": \"WanVideo T5 Text Encoder Loader\"\n }\n },\n \"16\": {\n \"inputs\": {\n \"positive_prompt\": \"猫咪的头部猛的向前一探\",\n \"negative_prompt\": \"色调艳丽，过曝，静态，细节模糊不清，字幕，风格，作品，画作，画面，静止，整体发灰，最差质量，低质量，JPEG压缩残留，丑陋的，残缺的，多余的手指，画得不好的手部，画得不好的脸部，畸形的，毁容的，形态畸形的肢体，手指融合，静止不动的画面，杂乱的背景，三条腿，背景人很多，倒着走\",\n \"force_offload\": true,\n \"use_disk_cache\": false,\n \"device\": \"gpu\",\n \"t5\": [\n \"11\",\n 0\n ]\n },\n \"class_type\": \"WanVideoTextEncode\",\n \"_meta\": {\n \"title\": \"WanVideo TextEncode\"\n }\n },\n \"28\": {\n \"inputs\": {\n \"enable_vae_tiling\": false,\n \"tile_x\": 272,\n \"tile_y\": 272,\n \"tile_stride_x\": 144,\n \"tile_stride_y\": 128,\n \"normalization\": \"default\",\n \"vae\": [\n \"38\",\n 0\n ],\n \"samples\": [\n \"90\",\n 0\n ]\n },\n \"class_type\": \"WanVideoDecode\",\n \"_meta\": {\n \"title\": \"WanVideo Decode\"\n }\n },\n \"38\": {\n \"inputs\": {\n \"model_name\": \"Wan2_1_VAE_bf16.safetensors\",\n \"precision\": \"bf16\"\n },\n \"class_type\": \"WanVideoVAELoader\",\n \"_meta\": {\n \"title\": \"WanVideo VAE Loader\"\n }\n },\n \"39\": {\n \"inputs\": {\n \"blocks_to_swap\": 20,\n \"offload_img_emb\": false,\n \"offload_txt_emb\": false,\n \"use_non_blocking\": false,\n \"vace_blocks_to_swap\": 1\n },\n \"class_type\": \"WanVideoBlockSwap\",\n \"_meta\": {\n \"title\": \"WanVideo Block Swap\"\n }\n },\n \"56\": {\n \"inputs\": {\n \"lora\": \"Wan21_I2V_14B_lightx2v_cfg_step_distill_lora_rank64.safetensors\",\n \"strength\": 3,\n \"low_mem_load\": true,\n \"merge_loras\": false\n },\n \"class_type\": \"WanVideoLoraSelect\",\n \"_meta\": {\n \"title\": \"WanVideo Lora Select\"\n }\n },\n \"89\": {\n \"inputs\": {\n \"width\": 832,\n \"height\": 480,\n \"num_frames\": 81,\n \"noise_aug_strength\": 0,\n \"start_latent_strength\": 1,\n \"end_latent_strength\": 1,\n \"force_offload\": true,\n \"fun_or_fl2v_model\": false,\n \"tiled_vae\": false,\n \"vae\": [\n \"38\",\n 0\n ]\n },\n \"class_type\": \"WanVideoImageToVideoEncode\",\n \"_meta\": {\n \"title\": \"WanVideo ImageToVideo Encode\"\n }\n },\n \"90\": {\n \"inputs\": {\n \"steps\": 8,\n \"cfg\": 1,\n \"shift\": 8,\n \"seed\": 652251640766870,\n \"force_offload\": true,\n \"scheduler\": \"unipc\",\n \"riflex_freq_index\": 0,\n \"denoise_strength\": 1,\n \"batched_cfg\": false,\n \"rope_function\": \"comfy\",\n \"start_step\": 4,\n \"end_step\": -1,\n \"image_embeds\": [\n \"89\",\n 0\n ],\n \"text_embeds\": [\n \"16\",\n 0\n ]\n },\n \"class_type\": \"WanVideoSampler\",\n \"_meta\": {\n \"title\": \"WanVideo Sampler\"\n }\n },\n \"150\": {\n \"inputs\": {\n \"lora\": [\n \"56\",\n 0\n ]\n },\n \"class_type\": \"WanVideoSetLoRAs\",\n \"_meta\": {\n \"title\": \"WanVideo Set LoRAs\"\n }\n }\n}"
49+
"description": "wan2.2是图像生成视频的模型，主要用来帮助用户根据图片加提示词来生成视频。",
50+
"content": "{\n \"11\": {\n \"inputs\": {\n \"model_name\": \"umt5-xxl-enc-bf16.safetensors\",\n \"precision\": \"bf16\",\n \"load_device\": \"offload_device\",\n \"quantization\": \"disabled\"\n },\n \"class_type\": \"LoadWanVideoT5TextEncoder\",\n \"_meta\": {\n \"title\": \"WanVideo T5 Text Encoder Loader\"\n }\n },\n \"16\": {\n \"inputs\": {\n \"positive_prompt\": \"猫咪的头部猛的向前一探\",\n \"negative_prompt\": \"色调艳丽，过曝，静态，细节模糊不清，字幕，风格，作品，画作，画面，静止，整体发灰，最差质量，低质量，JPEG压缩残留，丑陋的，残缺的，多余的手指，画得不好的手部，画得不好的脸部，畸形的，毁容的，形态畸形的肢体，手指融合，静止不动的画面，杂乱的背景，三条腿，背景人很多，倒着走\",\n \"force_offload\": true,\n \"use_disk_cache\": false,\n \"device\": \"gpu\",\n \"t5\": [\n \"11\",\n 0\n ]\n },\n \"class_type\": \"WanVideoTextEncode\",\n \"_meta\": {\n \"title\": \"WanVideo TextEncode\"\n }\n },\n \"28\": {\n \"inputs\": {\n \"enable_vae_tiling\": false,\n \"tile_x\": 272,\n \"tile_y\": 272,\n \"tile_stride_x\": 144,\n \"tile_stride_y\": 128,\n \"normalization\": \"default\",\n \"vae\": [\n \"38\",\n 0\n ],\n \"samples\": [\n \"90\",\n 0\n ]\n },\n \"class_type\": \"WanVideoDecode\",\n \"_meta\": {\n \"title\": \"WanVideo Decode\"\n }\n },\n \"38\": {\n \"inputs\": {\n \"model_name\": \"Wan2_1_VAE_bf16.safetensors\",\n \"precision\": \"bf16\"\n },\n \"class_type\": \"WanVideoVAELoader\",\n \"_meta\": {\n \"title\": \"WanVideo VAE Loader\"\n }\n },\n \"39\": {\n \"inputs\": {\n \"blocks_to_swap\": 20,\n \"offload_img_emb\": false,\n \"offload_txt_emb\": false,\n \"use_non_blocking\": false,\n \"vace_blocks_to_swap\": 1\n },\n \"class_type\": \"WanVideoBlockSwap\",\n \"_meta\": {\n \"title\": \"WanVideo Block Swap\"\n }\n },\n \"56\": {\n \"inputs\": {\n \"lora\": \"Wan21_I2V_14B_lightx2v_cfg_step_distill_lora_rank64.safetensors\",\n \"strength\": 3,\n \"low_mem_load\": true,\n \"merge_loras\": false\n },\n \"class_type\": \"WanVideoLoraSelect\",\n \"_meta\": {\n \"title\": \"WanVideo Lora Select\"\n }\n },\n \"89\": {\n \"inputs\": {\n \"width\": 832,\n \"height\": 480,\n \"num_frames\": 81,\n \"noise_aug_strength\": 0,\n \"start_latent_strength\": 1,\n \"end_latent_strength\": 1,\n \"force_offload\": true,\n \"fun_or_fl2v_model\": false,\n \"tiled_vae\": false,\n \"vae\": [\n \"38\",\n 0\n ]\n },\n \"class_type\": \"WanVideoImageToVideoEncode\",\n \"_meta\": {\n \"title\": \"WanVideo ImageToVideo Encode\"\n }\n },\n \"90\": {\n \"inputs\": {\n \"steps\": 8,\n \"cfg\": 1,\n \"shift\": 8,\n \"seed\": 652251640766870,\n \"force_offload\": true,\n \"scheduler\": \"unipc\",\n \"riflex_freq_index\": 0,\n \"denoise_strength\": 1,\n \"batched_cfg\": false,\n \"rope_function\": \"comfy\",\n \"start_step\": 4,\n \"end_step\": -1,\n \"image_embeds\": [\n \"89\",\n 0\n ],\n \"text_embeds\": [\n \"16\",\n 0\n ]\n },\n \"class_type\": \"WanVideoSampler\",\n \"_meta\": {\n \"title\": \"WanVideo Sampler\"\n }\n },\n \"150\": {\n \"inputs\": {\n \"lora\": [\n \"56\",\n 0\n ]\n },\n \"class_type\": \"WanVideoSetLoRAs\",\n \"_meta\": {\n \"title\": \"WanVideo Set LoRAs\"\n }\n }\n}"
5151
}
5252
]

0 commit comments

Comments
 (0)