This Model does not produce animated gif's automatically, it is a tool that I used to generate frames and then stitch them together to make gifs.
Currently working on Control Net Support and then also another section of the graph for turning Images into videos.
What I have noticed is that alot of ComfyUI's forget to mention that you need to install KJNodes for Get and Set but ComfyUI doesn't do it automatically.
I check CivitAI often, so feel free to ask questions in the comments, and please post what you generate!
Take the first Krystal image put drag it into your ComfyUI to generate the workflow, and it should work.
If you generate anything with it, my single request is that you please post it here. Civitai doesn't list tools and workflows automatically so unless you post it here it will never get tagged with Mooseflow and it would be nice to see how people are genning images with it
Description
Quality of life Changes:
RhthreeHotkey for travelling around the graph.
1 - Provided you have a 16:9 monitor, you should have access to all the important prompt info here
2 - Lora focus
3 - Lora trigger word focus
4 - Inpainting focus
5 - Refiners focus (becareful touching these if you are new to comfyui)
Metadata node so that Images contain metadata for Civitai.
It should notify you when a gen is done now, properly.
Cleaned it up a little bit, got all the weird stuff off.
FAQ
Comments (31)
What is workflow for?
NSFW mostly, but it’s also good for SFW.
@BoolMoose Usually i don t have making nswf content on pony and Illustrious. In description i see. That it can do Gifs by controlnet. So it's a workflow for making gifs? Or it s normal text to image workflow?
@KeMiliUs Sure I can explain, Im still working on the control net and gifs part, so far I did a bunch of work as a proof of concept to make a gif, but it really does require a bit of work. Took me 6 hours of messing around with it just to make that short gif via AI, next I will be adding control net nodes to help make its gen more strict. Its not like the other image to video gens. I made this workflow because those video models don't really do hentai very well, and they are also still in SD 1.5 which is very old and doesn't listen to prompts well. So I use this workflow to generate frame by frame and then stitch them all together. Still a work in progress but Im getting there
@BoolMoose Ok, i will add this workflow in bookmark and will check updates
What is the node used in “Model name for Img Metadata”?
I'll be honest, at first I just used a string node (lol), because I didn't really have enough time to search for a node that grabbed that automatically. But that is also like the easiest thing to add in Civitai, so I didn't really put much attention to it. If you know a node that grabs it automatically lmk!
@BoolMoose I can't find what's a custom node?
My comfy doesn't download anything by itself..... I downloaded everything manually.
@RedTag Oooooh I see, get ComfyManager (https://www.youtube.com/watch?v=5AHBRVaJf9k&t=26s) it's supposed to download everything automatically (except KJnodes for some reason), once you have COmfyManager the rest shoouuuuld be easier, but if you have any other problems let me know
What folder does a workflow go in?
Comgyui/User/Default/Workflow
or you can just drag the image onto the window (called the graph)
Hello! Your workflow and it's results on /slop/ were inspiring enough for me to try comfy for the first time but I'm afraid i dont have enough familiarity with it to use this properly, specifically in "Face" "Hand" "Eye" and "Foot" Refine nodes I'm missing those bbox detector models despite downloading the Subpack. Would you be willing to compile a folder of the things you used for these gens (The LoRas, VAE, Checkpoint) and share it? I don't want to ask for too much of your time but seeing yours and the other anon's results is quite something!
Firstly make sure you have manager installed, and it sounds like you do
Loras - For illust I almost never actually need lora, except with things it struggle with, then I will use a Lora, on /slop/ It was mostly the BallsDeep Lora found here on Civitai
I believe these are the BBOX models from Civitai, (I actually downloaded most of mine through Manager -> Model Manager -> Searched by name) but these should be the same one on Civitai for simplicity sakes but im not 100% sure
https://civitai.com/models/178518/eyeful-or-robust-eye-detection-for-adetailer-comfyui
I use Face-yolo but I just discovered this https://civitai.com/models/1076050/adetailer-anime-girl-face-segmentation (not sure if I should use the segm or bbox Ive yet to try it out), so Im probably going to use this to keep my males faceless, but use Face/yolo if you want to make male faces detailed
For hands I use Hands-yolo, found in the model manager and I cant find it on Civitai
For feet I use https://civitai.com/models/257904/adetailer-footyolov8xpt
Thanks for the question I never bothered to look up detailers on Civitai and I found people are making a ton of other detectors, will definitely be expanding the refiner blox in the next Mooseflow update
@BoolMoose Thank you so much for the expedient reply! I almost got it all working but I have a few remaining questions if you don't mind. On node #38 "Empty Latent Image Custom Preset" I'm afraid I don't know where to make the presets nor how important they are so I've deleted it and used the Node below it to produce the image size for the gen, if i shouldn't do this i'd like to know.
Secondly in the "Prompt Command Center" Node #432 "Poisitive Prompt" I can't edit the text in that box. It's most likely my own ignorance but I don't know how to change it. I did manage to gen a few things but to the standards I've seen and I'm hoping to make it there soon!
@chim Yeah thats no problem, its just a ease of use thing for me. If you want to add a folder in there yourself the directory is ComfyUI>Custom Nodes>KJNodes>replace the Custom Dimensions file with the file I added to the resources of the same name. Or you can write in your own usiong Custom Dimensions Example, as a starting point, just make sure the file is named custom_dimensions
@chim Oh also the Positive prompt is not editible, I use the little prompt builder on the left of it. If you dont want to use the prompt builder you can look to the right and a huge text box is there for freeform prompting in, delete the prompt builder and unpin the SetNode for the Positive prompt and connect the freeform prompt box to the SetNode and it should prompt as normal
Any updates on the ControlNet and Img to Video support? I've got extra a few hundred buzz to give you when it gets put up, since my own attempts at integrating them have been meh at best.
Also only suggestion: Add some of those Turn on/ Off shortcuts to click off specific parts of the workflow.
did u adjust the strength of the ballsdeep lora or did you increase the strength of the prompt e.g. deep penetration:1 - deep penetration:1.4?
Mix of both, the balls deep lora only goes so far then I had to switch over to mating pressing
The workflow is relatively easy to understand, I am just confused with the nodes that get other connections, took me a while to understand that part
But overall looks really good
Btw, you are still working on the gif part right? I wanna try animation so bad but all the workflows I've seen so far mess with my head, I don't understand how they work
How can i set Empty Latent Image Custom Presets?
First of all,Hello saw your Post of 20 hours ago Nice to see you back!❤️❤️❤️
Can you share some more animated gifs if possible? also are You updating this workflow for controlnet support??
Latest update of Comfyui seems to render this workflow w/ no organization or nodes connected.. Probably not related to the creator at all but is there a way to fix this? One of my staple workflows prior to the UI undoing the whole thing.
Trying to get this to work but it's still saying nodes are missing?
NovelAIToComfyPrompt
NAIPromptExtractor
I cant find these two nodes. They are likely to have been abandoned
@gw6120 Bro, it helped. Thanks!
Can i ask why you using 2 or 3 times the same node?
I mean you using LoraInfo node and the ComfyUI-YCYY-LoraInfo node, for what reason?Ther are both the same node with the same input's and outputs.And the github page to the NAIPromt is gone