Hunter69 v1
release date: 17 January 2023
Due to many DM requests on Discord, I decided to share version 1 of my model which is a merge of some popular models (forgot which ones) and 3 private ones (not from Civitai or any other public website - but with permission), which I have perfectionized into the ideal checkpoint/model for creating stunning real life looking female nude models.
Despite the possibility of being able to create cartoons/anime/drawings/males/objects, I have focussed a lot towards generating female (solo) nude art models similar to the style of that from MetArt & Domai.
I'm currently am working on v3, which is already a huge improvement over v1 after v2 became corrupted (in the sense of missing a few amount of bytes during transfer from server to server) and therefore being unable to load it.
When will v3 be released?
Most-likely around April 2022 or earlier, as I want to merge it with more embeddings as well as merge it with a yet to be privately released model trained completely by myself.
There won't be new versions of this particular model on Civitai, Hunter v3 will most-likely be released with a different name and of course on Civitai!
Updates
No longer using Patreon
Merges
I support creativity, if you plan to use this model for merges, you do not have to credit me, that's up to you. But thanks in advance for doing so.
Support my efforts
Want to support my efforts, which take a lot of time, testing, plotting, comparing, rendering, generating you can simply buy me a coffee and help me reach my goals.
You will not only help support me with a brand new videocard I'm saving for, which will make training of my new future models from scratch (no merging) much faster, but you'll also get access to HQ (1024/1280/1368px) images I have created myself and more!
Notes
This model requires vae-ft-mse-840000-ema-pruned
Despite me using my model mostly for 21-30 year old Asians females, it does work for any ethnicity and/or age as shown in the gallery of X/Y plotting
Not enough Examples? Visit my site here (be aware that all images shown are based on Hunter69 v1, and I have stopped updating)
Description
There won't be new versions of this particular model on Civitai, Hunter v3 will most-likely be released with a different name.
FAQ
Comments (29)
You need to have sample prompts with your images, to encourage people to test your model
Yeah, Hunter's right that he doesn't need anything, but OP is right that I'm not going to bother with your checkpoint without prompts. Most likely they were just done with another model and you've rammed some half-baked merge onto Civit. It's either lazy or selfish, dude. Chill, share.
If i would be selfish, I wouldn't post anything here at all.
besides, 1 image with prompt is more than enough
UPDATE: Apologies, I didn't notice that my prompt wasn't saved earlier, I have just saved it for the first image, as that one was most requested.
@Hunter69 Not a convincing argument when you have a patreon advertisement in the model description.
@wewewew I don't see anything wrong with that?
So how should this model be prompted? Kind of hard to guess what works.
- Make use of what makes most sense
- Try X/Y Plotting to find better prompts
- I sometimes posts prompts as well as X/Y plotting results of my model on Discord
- Try your current saved prompts from your current model(s), and go from there to get better results
This model is a demonstration of what have been generated with my model
note: there is 1 image thought with a prompt, that might help as well?
UPDATE: Apologies, I didn't notice that my prompt wasn't saved earlier, I have just saved it for the first image, as that one was most requested.
Fixed permissions
your patron link is incorrect.
It's being reviewed to check if its really AI or real.... :(
At this moment the link doesn't work, but hopefully it will soon be approved and enabled
Safetensor model?
Edit:ty!
Please check again @ewe
Just replaced the cpkt with safetensors ;)
My new favorite model.
Thanks alot for the upload, appreciate it.
I've briefly tried it but my results are nothing like your beautiful and incredible examples here: https://hunter69.com/
Any chance you could tell us which sampling method, CFG Scale etc you generally use to achieve these results?
Thanks.
try higher resolutions such as 768x768, cfg 9.5 for example, and perhaps 42 steps would be a good start
@Hunter69 i mean, if i put like, 150 samples, and 30 cfg, would it be worse or better?, which one and why
@helloweek2 keep CFG please @ 8 to 12 and please keep steps between 30 and 50 CFG, anyhting under or below this, will result in bad images (when your goal is realistic photograpy like images)
add some more sample propts
safe tensor version would be appreciated
Just updated the file, it is now a .safetensor file
i never tried 'Scandinavian' to force white skin... other were move obvious, although 'African' can affect entire picture ;(
you can also try pale skin, or (pale skin:1.2) and so on ;)
I'm using automatic1111 and get those images.
please use the VAE that I've mentioned in the description, and it should fix this problem ;)
Let me know how it goes
@Hunter69 Thank you! it works fine.
nice
Just one comment, to get more popularity, would be useful share the prompts you use to get good results, because only viewing your images.....well that is not enough....at least one example
just a comment