User Guide
1.This workflow uses the original Qwen-Image-Edit2511 model combined with the multiple-angles LoRA model to generate multi-angle images, or allows manual angle control for creating images from different perspectives. It can be used for storyboard creation and LoRA material library development.
2.Switch between multi-angle and single-angle control via a toggle—only one function can be used at a time.
3.Pre-annotated multi-angle prompts are provided for selection. Simply copy and paste the required prompts; the number of pasted lines corresponds to the number of images generated.
4.Single-angle control is visualized through the ComfyUI-qwenmultiangle node. Download link: https://github.com/jtydhr88/ComfyUI-qwenmultiangle
5.Adjust resolution settings in the Parameter Selection Area.
6.An optional AIO model loader is included. However, tests show that the AIO model has lower character consistency compared to the original 2511 model. It is provided for specific use cases, with version 18.1 (the most effective one) included instead of the latest updates.
7.Fine-tune other parameters based on your specific requirements.
8.The angles in this workflow are controlled via prompts and are still subject to the limitations of prompt engineering. For precise angle control, the Gaussian splatting feature is required.
Qwen-Image-Edit-2511-Multiple-Angles-LoRA：https://civarchive.com/models/2300308/qwen-image-edit-2511-multiple-angles-lora
This workflow is available for free online testing. You can download it for local use if you find it useful.
URL: https://www.runninghub.ai/post/2018253621518929922/?inviteCode=rh-v1443
Description
FAQ
Comments (8)
Hello, do you know if the same exists for Flux2Klein? Thank you 🙌
AFAIK a multiple angles LoRA does not yet exist for Klein - could be wrong.
Flux2Klein has a similar workflow, but it relies solely on the model’s own capabilities and prompts, and I don’t think the results are very good.
The core of my workflow is the LoRA: qwen-image-edit-2511-multiple-angles — it’s what makes the output really stand out. Plus, Qwen Edit keeps character consistency really well.
Flux2Klein is decent, but I still think it’s not as good as Qwen. That said, Flux2Klein has a strong foundation, and as the community builds more around it, better models will definitely come along.
Wow very powerful results.
Thanks for your support!
Hello, I'm not getting the same results from this workflow. I only get 1 slightly different image when I select for multiple images. Do you have a video tutorial on how this works?
Unfortunately, there isn’t a video tutorial for this workflow available just yet.
To resolve the issue, first double-check that your prompt input node is configured exactly the same as the one in my workflow. This node is designed to generate one image per line of prompt you input — so the number of lines in your prompt will directly equal the number of images it outputs.
If you’re only getting a single image back, first confirm that your prompts are properly split into separate, individual lines. You’ll also want to verify that you’re using the exact same node version as the one included in my workflow.
The downloadable workflow and the workflow used for the example images are not the same. One uses the PromptLine node to handle multi-line prompts. The other just handles single prompts. I added a PromptLine node to the simpler workflow behind the download link. Connect the "string" output to the "prompt" input of your text encoder.