Hello world
Let's get real
It's been a while since the last update, and this iteration is quite different than the last. That's not to say it is necessarily groundbreaking compared to anything else on this website (it's a merge after all), but I've tried to focus on a couple of very specific things:
A high amount of realism without any use of LoRAs or embeddings.
Having a good basis provides a lot of room to go crazy with those later.
Functional with a limited/small negative prompt.
Similarly, a small negative prompt often makes for the most realistic outputs. It is always possible to write your novel, or add embeddings in there along the way.
Consistency in poses / hands / other features within these constraints.
Decent control over light and darkness through prompt alone
There are great LoRAs out there for lighting / noise offset, but there is already some great functionality by default.
Responsiveness to the wildest lists of wildcards ChatGPT has to offer.
I like to have random lists of locations, time periods, clothing/materials, concepts, all generated (and embellished) by your local GPT.
Recommended VAE (not included): vae-ft-mse-840000-ema-pruned
-High base resolutions are advised: such as 640x896 for a portrait format
-If possible use hires.fix or a similar implementation: a denoise of 0.3~0.5 will do well in most situations various ESRGAN upscalers give good results. x1_ITF_SkinDiffDetail_Lite_v1 is the one used in all examples here, but there are other great options.
-Don't feel obligated to use time consuming samplers such as the one in most examples. I was suffering from a short bout of SDE Karras addiction when generating the examples, but I have since come back to 2M Karras @ 15~20 steps, which is usually my baseline and more than fine on its own (even with hires.fix).
*Note that some of the example prompts can generate nudity, prompt against this to prevent it. For actual examples of nudity you can check out the album uploaded separately.
**The model in the metadata is QGO-3002a, which is the same as QGO-PromptingReal-v1
Description
This merge involves aaacups to have more variation in breast sizes, I personally have a wildcard that has the terms "big, huge, massive, giant small, tiny, flat, little, aaa" in it, followed by a "breasts, tits, boobs, chest" wildcard.
It uses RPGv4 instead of v3. It also involves Latex to get some of the bondage model back in as well..
FAQ
Comments (28)
Hi guys, does anybody know how this model behaves in comparison with Realistic vision regearding the skin texture? TY
Could you share your wildcard setup? This model is absolutely amazing!!
Hi, thanks! This is my wildcard folder https://www.mediafire.com/file/9342ic4y4ih7dh6/wildcards.zip/file It's not organized well, and a lot was quickly made on the spot. But it will allow you to use the templates from the prompts I suppose.
@qego Hi ! Love your work, do you have discord ? I'm new in SD and i like it so much already, but i need your help to install this model with all the extensions you used, sorry if my english is bad (i'm French ahah). If you're OK, send me a message to my discord : Mélih#1286 Thanks bro!
Which VAE do you suggest me to use? :)
Hi! I use the standard vae-ft-mse-840000-ema-pruned https://huggingface.co/stabilityai/sd-vae-ft-mse-original/tree/main
I have not gone on different VAE adventures yet :)
@qego Thank you!
been using this model for like a week and i love it.
I was wondering on how to get the same stuff as you? For the qgo-10a model, i used the pruned vae and all the same stuff for the "photo portrait of a Italian woman with slicked back hair" but i didnt get the same results.
Please prune the model, it's way too large
possible to include inpainting version for this and upcoming releases?
i think the safetensors for the cum model is broken. using the model results in broken results, and merging results in even worse results. im unwilling to try the non-safetensors version because i don't trust ckpt.
i don't have these issues with other people's models.
model
merged model.
These results occur when a VAE is required, but not being used; try applying the standard VAE.
@phreakheaven299 i see. i was using the ft_mse vae. i will try later without it.
Great examples, but my experience is disappointing.
I couldn't get anything close or even similar to the examples, although I used the same prompt string, the same number of steps, and the same seed. The result turned out to be completely different in terms of colors, light / shadow, skin texture, and plot (2 dressed Asians next to the TV and something like that). I use DiffusionBee and as far as I know I can't use external VAE or LORA, although I could be wrong because I'm still new to this. Any help is welcome.
I used qgo10b_qgo10bFixFixedCLIP.ckpt model
qgo10b seems to be a little off, the previous one (qgo10a) felt more natural and easy, try to play with negative prompts, seems to be more responsive to them
looking at the prompts he seems to use lora and embeddings in positive and negative ones so yeah even if you copy paste all prompts,seeds,sampler etc. it will not have the same image or close to it
DiffusionBee uses a Tensorflow back end. I'll bet these were made with Automatic1111 or another PyTorch-based app. And if you're using Bee, are you on a Mac? Forget matching PC gens. I have a PC and Mac side-by-side both running Auto1111 with same models, params, prompts, etc. and I can't get gens to match. Guys are upgrading their graphics cards and finding they can't match their own old gens.
Really a great model ! Can you upload a smaller 2gb pruned fp16 version ?
Great job! Can you tell me how to make the model "Base10+Add11Phot-40-60"?
Great model but huge. Can you add, if possible, a smaller 2gb pruned fp16 no-baked-vae safetensor version for the v1.0b-fixed ?
Thanks for your work! But I can't generate small/tiny/little breasts using this model.😢
Hi, your artworks is really very nice, our company “LiveMe” would like to cooperate with you for AIGC model training, do you have email or other contact information?
How do I find qgo-10a?
right below the name and then the tags you can chose the version. click it, then click download on the right
hi, is it recommended to download the CLIP fixed version instead of the normal 10b?
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.