Update April 28th Added version V4 pruned model.
A model with professional RPG industry artists in mind. Those working in video games, board and tabletop games as well as concept art and book covers should get good use from this model. The new version, V4, packs in more training for creatures and RPG subjects including a more illustrative style.
Detailed tutorial on how I get the results in the preview images.
Check here if you're having trouble getting the same results. Initial generation size and VAE usage is key.
You can prompt any style you need with these models, but the default aesthetic is listed for each of the models in this handy list.
Different models available, check the blue tabs above the images up top:
Stable Diffusion 1.5 (512) versions:
V4 inpaint Inpainting version of V4 that's good for outpainting.
V4+VAE Same as V4 but with the added convenience of having a preset VAE baked in so you don't need to select that each time. Saves on vram usage and possible NaN errors. Speeds up workflow if that's the VAE you're going to use anyway.
V4 New aesthetic added to make it a bit more illustrative. More training for better coherency for all things RPG.
V3+VAE Same as V3 but with the added convenience of having a preset VAE baked in so you don't need to select that each time. Saves on vram usage and possible NaN errors. Speeds up workflow if that's the VAE you're going to use anyway.
V3 Stronger painterly style. High contrast and sharpness. Even more RPG knowledge.
V3 inpaint Inpainting version of V3 that's good for outpainting.
V2 Stronger painterly style. High contrast and sharpness. More RPG knowledge.
V2 offset Noise Offset added making more contrast and bringing the model back to photoreal.
V2 Art Trained model. Very artsy. Strongest painterly style. Less details and bigger brush strokes to mimic digital painting style pre-AI.
V2 inpaint Inpainting version of V2 that's good for outpainting.
V1 Smoother renders with least painterly effect.
V1 inpaint Inpainting version of V1 that's good for outpainting.
Stable Diffusion 2.1 (768) versions:
SD 2.1 768 V1 Strong painterly style, very coherent with hands and objects. Higher native resolution and detail. Not good for nudity.
Not as effective, but here's the LoRA if you like to use that instead:
A to Zovya RPG Artist's Tools LoRA
Do you have requests? I've been putting in many more hours lately with this. That's my problem, not yours. But if you'd like to tip me, buy me a beer. Beer encourages me to ignore work and make AI models instead. Tip and make a request. I'll give it a shot if I can. Here at Ko-Fi
Description
Stable Diffusion 1.5 (512 x 512)
A-Zovya RPG Artist Tools V2
Preferred model for inpainting, or outpainting.
VAE recommended vae-ft-mse-840000-ema-pruned.ckpt
FAQ
Comments (28)
I just wanted to point out that the SD2.1-768 v1 inpainting checkpoint shows 1.5 as base model, when it should be 2.1
fixed it. thanks for pointing that out.
The amazing thing about Zovya is that she is a genuine, bona fide artist.
I can feel something alive inside this machine, this AI model. As if her brush was moving and painting the fine detail on my dresses, armors, and designs when I press the Generate button...
Some call this the "Ghost in the Shell".
I can tell you, it's not a ghost, it's the beating heart of a generous and talented individual committed to bringing YOU high quality tools for FREE.
If you are on this page I can assure you that you are in a very good place. Make it even better by pressing the download button, by adding this model to your favorites, and by leaving a kind review that will help the creator.
Best,
EDG
Best comment of 2023.
Hear Hear!
How come whenever I make an image it has a blurry dark look, even for deforum its like blurry compared to images provided below. I do it on all resolutions too.
Like it looks like a watercolor painting if that makes sense
I had pretty much the same issues that you describe, until I followed the steps outlined at the link that was given in the description, that might help?
Detailed tutorial on how I get the results in the preview images.
@TheEnigma because there is no VAE, you have to use one.
@HiP_fr Where can I use a VAE, I know that I can put them in a folder but how do I use it? Is it in the settings tab?
@TheEnigma hi... (assuming you are using Automatic1111) a VAE can be dropped in your [SD path]\models\VAE folder, next start up your Web GUI and head to 'Settings', look up 'Stable diffusion', and find the setting 'SD VAE', and select yours from the drop down list box. (if none show up, press the refresh button on the right side of the drop down box. After that, click the 'Apply settings', and restart Stable diffusion (not just the UI, but the .bat file), and you should be good to go. There's many VAE's available (with various effects on your outcome when generating images) but the one I use most is vae-ft-mse-840000-ema-pruned, which you can download on hugginface (https://huggingface.co/stabilityai/sd-vae-ft-mse-original/blob/main/vae-ft-mse-840000-ema-pruned.ckpt), or right here on civit.ai: https://civitai.com/models/76118/vae-ft-mse-840000-ema-pruned.
@ceeveranda177 Thank you so much, this fixed the problem and your detailed steps were exactly what I was looking for. I really appreciate you helping out!
You should specify clearly that we need the VAE to achieve your results, instead of just pointing to the other page.
I do. on each version I list the recommended VAE.
@Zovya I didn't see it anywhere on this page description.
but i'm not saying it agressively. I just wanted it to be as friction free as possible to enjoy your amazing work here, even for new people coming in everyday.
@HiP_fr thank you. I appreciate ALL feedback
So far I'm loving your V3 model! I have a hard time reproducing your preview images though: When I use your prompts I get something different - the changes often quite noticable (e.g. a castle in the background is missing, a dress is green instead of blue, different pose etc.). Any thoughts what the reason might be? (I'm using the VAE vae-ft-mse-840000-ema-pruned.ckpt as you said in that linked article, but maybe I'm doing something wrong - I'm fairly new to Stable Diffusion).
Another question: I noticed that you use the tag "zrpgstyle" in some of your prompts - what is the purpose of that? (Using that tag changes the image a lot, but not the style.)
Thanks in advance!
I usually like to go crazy with the first preview image. In this case there's a lot of controlnet and inpainting to get the v3 logo in there. All the other images are straight txt2img. The zrpgstyle was a trigger for the style in previous versions. It's not as strong in this version.
@Zovya Thanks for the quick reply :) I still don't understand why I get different results with the rest of the text2img pictures, but maybe that's a bug (I read that batch size affects the output, even when using the same seed) or maybe it's RNG related (do you use GPU or CPU?). At the end of the day it's not that important though, the results are excellent nonetheless.
Another quick question if you don't mind: Some of the preview pictures seem to have been generated with your V3 model + the RPG-Lora you made. I tried that too (copied the prompt) but I only get a garbled mess as a result. Why?
@mike92011 hard for me to know what your problem is without seeing your results. There are many factors that could contribute. I know meta data doesn't show my initial image size which is 512x768 for most images. Then I use hires fix to get the full size. I describe that in the tutorial I linked though. If you're still having troubles, ping me on the civitai discord and we can work it out.
@Zovya Thanks for the offer :) I will contact you once I have made some screenshots.
@Zaratustra Well, I did it for testing purposes and I was just curious why I couldn't reproduce the images because I could do that with the preview images of other models, that's all. As I said, it's not so important to reproduce these images exactly, because it's more about the art style and what the model can achieve.
do the civit models get updated automatically or do i have to come and get them again?
if there's a newer version, you'll need to get that version if you want it. but older versions have their place too, it's why we don't delete them
Been having lots of fun with V3. Really appreciate the work you've done with it. Kinda bummed I can't get the upscaler to work with 8GB Vram :(
That said I do have a question though: does v3 support bird races, or are there particular prompts I should use? I saw that you've got dragonborns working, but it seems to be having trouble doing birds.
yeah works pretty well. I just prompted Kenku and Aarakocra creature with wings
https://civitai.com/images/1008937?modelVersionId=87886&prioritizedUserIds=48396&period=AllTime&sort=Most%20Reactions&limit=20
check the controlnet tiles as an option for upscaling
@Zovya sweet thanks, I guess I'll have to try some more