if-ai / ComfyUI-IF_AI_tools

ComfyUI-IF_AI_tools is a set of custom nodes for ComfyUI that allows you to generate prompts using a local Large Language Model (LLM) via Ollama. This tool enables you to enhance your image generation workflow by leveraging the power of language models.
https://ko-fi.com/impactframes
475 stars 35 forks source link

How can I use the content generated by dolphin_llama3_omost in WebUI? #99

Open Amazon90 opened 3 weeks ago

Amazon90 commented 3 weeks ago

I find the use of ComfyUI nodes too difficult. I would like to ask the author if I can directly use the dolphin_llama3_omost model and copy the results generated by Ollama into the prompt. After testing it myself, I found that SD Forge did generate an image, but I'm not sure if this approach is correct. Should I avoid copying all of the content generated by Ollama, and instead extract a specific part of it?

ollama run impactframes/dolphin_llama3_omost

draw a girl


# Initialize the canvas
canvas = canvas()

Set a global description for the image

canvas.set_global_description('A cute girl in a summer dress with a hat, standing by a window on an overcast day.')

Add the girl wearing a summer dress and a hat.

canvas.add( location=1, area=0.3, offset=0.05, distance_to_viewer=1.5, description='The girl wearing a summer dress and a hat.', detailed_descriptions=[ 'A cute young girl is standing by a window on an overcast day, dressed in a light summer dress with floral patterns.', "She has a bright smile on her face as she gazes out the window.", 'Her hair is neatly tied back into a ponytail, and she wears a wide-brimmed sun hat that shields her from the soft sunlight filtering through the curtains.', ], tags='girl, summer dress, hat, smiling, by window, overcast day, floral patterns, ponytail, cute girl, young girl, bright smile, windowpane, light fabric, cheerful expression', )

Add a potted plant next to the girl.

canvas.add( location=2, area=0.15, offset=0.08, distance_to_viewer=1.5, description='A potted plant next to the girl.', detailed_descriptions=[ 'Beside the girl is a small, potted plant in a simple white pot.', "The plant's leaves are lush and green, adding a touch of nature to the scene.", 'It stands on a small wooden table with a smooth surface.', "The plant seems healthy and vibrant, complementing the girl's cheerful appearance.", ], tags='potted plant, green leaves, white pot, small table, wooden table, nature element, lush plant, indoor plant, complementary colors, healthy plant', )

Add a window frame with curtains.

canvas.add( location=3, area=0.2, offset=0.1, distance_to_viewer=1.5, description='A window frame with curtains.', detailed_descriptions=[ 'The window the girl is standing by has an old-fashioned wooden frame, painted in a soft cream color.', 'The windowsill is lined with white lace curtains that gently sway as if there's a light breeze coming through.', "These curtains filter the natural light, casting a gentle glow over the girl and her surroundings.", ], tags='window frame, wooden frame, old-fashioned window, cream paint, windowsill, white lace curtains, swaying curtains, natural light, gentle glow', )

Add a small table with a cup of tea.

canvas.add( location=4, area=0.1, offset=0.12, distance_to_viewer=1.5, description='A small table with a cup of tea.', detailed_descriptions=[ "In front of the girl, on the wooden floor next to her, there is a small round table with a delicate porcelain teacup and saucer resting on top.", 'The teacup holds freshly brewed tea that gives off steam, suggesting warmth and comfort.', 'The cup has elegant floral designs around its rim, matching the summer dress worn by the girl.', ], tags='small table, wooden floor, round table, porcelain teacup, delicate saucer, freshly brewed tea, steam rising, floral design, elegant cup', )

Add a bookshelf with books and a small picture frame.

canvas.add( location=5, area=0.2, offset=0.15, distance_to_viewer=1.8, description='A bookshelf with books and a small picture frame.', detailed_descriptions=[ 'To the right of the girl is an elegant wooden bookshelf filled with various books, neatly arranged in descending order by size.', "There's also a small picture frame hanging on one of the shelves, featuring a photograph of the girl's family.", 'The bookshelf adds a touch of sophistication to the room, making it feel cozy and lived-in.', ], tags='bookshelf, wooden shelf, books arranged, descending order, small picture frame, family photo, elegant bookshelf, sophistication, cozy atmosphere', )


![1](https://github.com/user-attachments/assets/12804456-4f33-4421-8805-d9bdda1a99dd)
![00001-174641794](https://github.com/user-attachments/assets/32dfc80a-3b6b-4db4-9bfa-908322703194)
if-ai commented 3 weeks ago

No is not correct because what the model does is generate regions and assign prompts to it with it then there are a couple of subsequent steps where a script takes the code defining the masks, regions colors etc.. and drives the conditioning on those areas. The guy that made forge is the same that made Omost so I am surprised if there is no way to use it already in forge?

Amazon90 commented 3 weeks ago

Yes, I also regret that I couldn't find an extension to use omost in Forge. However, after sending the content I wanted to paint to dolphin_llama3_omost, the results it provided were better than what I could generate with the prompt I sent.

if-ai commented 3 weeks ago

Eventually I want to bring this same tools to IFPromptMkr my extension for A1111 & Forge but the work here is too much and I can't do that until I least finish some of the features for IF_AI_tools

Amazon90 commented 3 weeks ago

If you could provide workflows for dolphin_llama3_omost and flux, that would be great. I’m not very familiar with using nodes in ComfyUI, so I can’t create the workflows myself.