A big thanks to civit77899 and AFD_0 for this V5 model and all models going forward
Description
This is my first model to publish. I'm in the learning stage
FAQ
Comments (12)
Thought I would offer some encouragement...this model works well. It's very consistent in output, seems to follow prompts well, and offers nice detail in both subjects and backgrounds. It seems to get a little confused when there are more subjects or more action - but that's a lot of models. Great job for a first merge.
Thank you for your input, inputs like this helps towards the next build
If this is your first checkpoint, you're either very skilled, very fortunate, or both. If you examine my images you'll see realistic accurate results that follow prompts to a good depth - 3 or 4 levels - and are very pleasing.
Terrific stuff and very well done. You get a Gold Star from me. Work on hand anatomy would be a good idea though.
Thank you. My focus on this model was prompting. Most of my images were from other peoples prompts, then I would look at how well their image matched their prompt vs mine. That was how I would decide if I wanted to use that model. My focus now has been looking at the hands and feet, also thinking about merging already merged models vs trained models.
I really do appreciate your feedback, thank you
I like your style of images, very artistic. Interesting to see a image of someone's personality. You're very creative
@jrrytng467 Thanks. Glad you like them. I'm a typical graphic designer with a butterfly brain!
This is an amazing merge! Can throw a wall of text at it followed by a ton of negatives and a dozen LoRAs stacked on top and somehow both prompt and LoRA adherence still work as expected, all while keeping peak image quality and realism for a modern XL-based model. Lighting and contrast can be a bit off sometimes, but quite often, it's near-ideal as well.
The results can be quite stunning and look a lot like MoP v5.1 XL Non-DMD by @GBRX with maybe a bit of extra Pony, but the prompting feels very Illu for some reason, which is nice. Very impressive for a first test merge. Think this will be a new favorite of mine that I'll be revisiting again (and/or future revisions), thanks for sharing!
Thank you. I have begun working on another one for better hands and feet while holding on to good prompting. Im also leaning towards models that have been trained and not merged. Im learning as I go, I do a ton of testing but the real test is everyone one here posting images and feed back. I do welcome neg. feed back for the chance to improve my models. I also love being able to give back.
Have you tried merging anything with it? Im curious about the results. Because I'm going through a learning stage I've been having issues with other models. After talking with ChatGPT, I'm going to have it teach me ComfyUI. With the base of my Test V1 I found out the author of it had merged clips by accident, i found the before and after models used in his spit. Instead of trying to fix mine, I'm going to try and recreate his model in steps and make sure that other clip doesn't follow. I love my model as a stand alone but I want it to be a good base model that myself or other can use it as a solid base model
I still enjoy using this model, I'm still realizing things I can do with it. I'm creating images to train a general model that would be a good all around model without extra things built into it so it would make a good base model for mergin