wan2.2+dasiwa+High dynamic range.
Model Source:https://civarchive.com/models/1981116/dasiwa-wan-22-i2v-14b-or-lightspeed-or-safetensors
🔽 📱 For Beginners (Easy AI App - Phone Friendly!)🔽
AI APP:
https://www.runninghub.ai/ai-detail/1993937472344227842/?inviteCode=rh-v1291
model&Lora&plug-in:
Download link for Qwen3VL NSFW Version:
https://pan.quark.cn/s/71503112e6e8
Download link for local Model & Loras:
https://pan.quark.cn/s/dbace2536daf
Download link for ComfyUI_Qwen3-VL-Instruct-abliterated: https://github.com/T8mars/ComfyUI_Qwen3-VL-Instruct-abliterated
Download link for ComfyUI-PainterI2V:
https://github.com/princepainter/ComfyUI-PainterI2V.git
✨ Fan Perks ✨ Click the avatar in the top right corner → Invitation Code → Enter code: rhv1291 🎁 Claim 1,000 FREE RH Coins immediately! 💰 Plus, get an extra 100 coins every day just for logging in.
Description
FAQ
Comments (5)
All your work is wonderful! And the way you present it is so wonderful.
I'm having a problem with Qwen3 VQA. You have the uncensored version by default. But it's not in the drop-down list, even after I downloaded it. The model doesn't recognize it. THAT'S the problem!
You'll need to download the full model repository from this link and replace the original Qwen3-VL model files to see the results..https://huggingface.co/huihui-ai/Huihui-Qwen3-VL-8B-Instruct-abliterated/tree/main
Naturally, I loaded both 8B and 4B. BUT VQA doesn't see them. I'm not that knowledgeable about this stuff. Instead, it loads regular instructions because the uncensored ones aren't written into this module. If you could add it to the patch, the uncensored version, so it would download it. That would be wonderful. I've been struggling for a long time and still haven't managed to get anything to work.
@bionovafood863 Are you sure you downloaded the node for this repository?https://github.com/T8mars/ComfyUI_Qwen3-VL-Instruct-abliterated The nodes in this repository have been modified and can load the "huihui" model.
You can also check out my latest update for Qwen 3.5, which includes an unrestricted version of the model and runs even faster. However, there are two scenarios. First, if you’re running locally, you can download my wan2.2 remix workflow, which already includes the qwen3.5 component. It performs exceptionally well and can generate NSFW content even when running locally. The second scenario is if you insist on using the workflow described in this post; in that case, you’ll need to know how to integrate the reverse-engineering component into this workflow.
@bionovafood863 This is the workflow page integrated with the qwen 3.5 auto-suggest feature:https://civitai.com/models/2466943/wan22remix-i2v-high-dynamic-rangeandauto-prompt