For the best experience, use it together with a futanari friendly checkpoint like Futanari Factor
This is an "improved" version of the FutaVeinyV6 that has more penis and testicles sizes. The adapter was trained using LoKr and was tested together with models that are already able to draw futanaris without much effort like abyssFuta, bb95-furry-mix, fluffyrock and their merges.
How to use
Edit the "SIZE" to the size you want to use.
Penis sizes:
small penis (WIP)
penis
large penis
huge penis
gigantic penis
Testicles sizes:
testicles
large testicles
huge testicles
gigantic testicles (WIP)
If the size slider doesn't seem to be working well with your prompt, try adding some weight to the token:
futanari, veiny penis, (gigantic penis:1.4), (huge testicles:1.2)Optional:
You can add "erection" or "flaccid" to your positive or negative prompts to control the dong status.
(WIP) If you like some saggers, add "sagging testicles" to your positive prompt.
(WIP) Add "foreskin, phimosis" to your positive or negative prompts to change the dong sheath.
Your feedback is appreciated! Its going to help a lot with tracking what checkpoints are working the best to produce the best dongs and what I need to change in the next iterations of this lora.
Development notes:
The training images were tagged using conv2 and swinv2 at the same time and then merged together and the duplicate tags removed.
Every caption was revised by me and the concept tokens were reordered to be used with keep_n_tokens.
Every image was cleaned of signatures and watermarks using lama cleaner and photoshop.
Censored images were redraw using deepcreampy and then manually retouched using manga studio.
5%~10% of the dataset is reserved for AI generated images made by me that were upscaled and manually retouched using manga studio and a drawing tablet. It helps to augment some concepts that don't have enough good pictures and achieve that specific futanari aesthetic I'm looking for.
I took 1 month to assemble, clean, redraw and tag the dataset, but the training and testing process is even more time consuming. To help with that, I'm also trying to code a tool to help me evaluate the testing grids, find the best epochs and track what worked or not between batches.
Description
This is an alpha version
Link to download all the epochs
High res version of the x/y grid
The last epoch seems to perform the best at bleeding art style, proportions and deliver the trained concepts. Leave a comment if you want need any other epoch uploaded.
Edit the "SIZE" to the size you want to use.
Penis sizes:
small penis (WIP)
penis
large penis
huge penis
gigantic penis
Testicles sizes:
testicles
large testicles
huge testicles
gigantic testicles (WIP)
If the size slider doesn't seem to be working well with your prompt, try adding some weight to the token:
futanari, veiny penis, (gigantic penis:1.4), (huge testicles:1.2)Optional:
You can add "erection" or "flaccid" to your positive or negative prompts to control the dong status.
(WIP) If you like some saggers, add "sagging testicles" to your positive prompt.
(WIP) Add "foreskin, phimosis" to your positive or negative prompts to change the dong sheath.
Changelog:
Replaced the last epoch from the download files with epoch 22 for better compatibility with the Futanari Factor model
FAQ
Comments (15)
How do you add BREAK into your prompts? I know its with Regional Prompter but how do you mask or do you even mask ?
I'm not using regional prompter in the featured pictures. The best way of explaining the effects of BREAK in your prompt would be:
- CLIP can only take 75 tokens at time
- to not truncate if it exceeded the limits, your prompt is segmented in chunks of 75 tokens, merged again and then sent to CLIP
- the first tokens in each chunk seem to have more effect in the final image
- tokens related to the same subject or part of the image composition seem to work best if chunked together
With all that info in mind, you can use BREAK to manually manage part of the chunking process and group related tokens together to increase the chances of getting a good image. In most of my prompts I use different chunks for the character, background, media type (traditional media, photo, render, etc.) and one last chunk for grouping my loras/lycoris. For instance:
```
1girl, long hair, futanari, looking at viewer,
BREAK
indoors, bathroom,
BREAK
traditional media,
BREAK
<lyco:FutanariConcept_LoKr_alpha:1.0>
```
Where are abyssFuta and the others mentioned available?
BB95
https://civitai.com/models/17649/bb95-furry-mix
Crosskemono
https://civitai.com/models/11888/crosskemonofurrymodelandhumanmodel
https://civitai.com/models/11888?modelVersionId=32830
Fluffyrock
https://huggingface.co/lodestones/furryrock-model-safetensors/tree/main
Monstercock
https://civitai.com/models/4165/monstercock
AbyssFuta
The original model was archived and deleted from civitai for no clear reason. There's no way of sharing it directly in here for now. The furry models are the ones doing most of the work anyway. You can merge some of the models I listed and see if you can reach similar results without using it.
Just for reference, counterfeitV30 and based64 seem to take futanari loras a bit better than most checkpoints and should be quite good for diluting the furry models if you're planning to merge a new formula.
@ckmlai Thanks for the response. Gonna try them.
@RedSkyAI you can now use this mix or follow the mixing steps and do your own https://civitai.com/models/110541/futanari-factor
What’s the chances this would work on men/photos/non anime?
It's probably not going to work well since all the dataset was entirely composed of anime style pictures. With some luck, the earlier epochs that I linked at description could work. In the worst case, using lora block weight or training a new version using realistic images could make it work.
Could you please make the lora version?
This dylora version should work as a lora
Hi, thanks, but they are lycoris right? Can we mix it to new check points?
@razzler It's classified as lycoris, but I think you can just place inside your lora folder and use it normally since I didn't use convolution ranks while training, defaulting for a resizable regular lora. Also, soon you will be able to use them normally, without any extension, by simply updating your webui. It's a new feature in the dev branch
Regarding compatibility with checkpoints and mixing, you should be able to do it normally. I recommend using the Futanari Factor that I released few weeks ago since it has more built in futanari knowledge than most models out there
@ckmlai sorry i mean i i can't mix the lycoris into checkpoints using extension like supermerger. Perhaps there's a new extension where we can mix lycoris to chekpoints? Anyway, your futanari-factor is awesome! It is great for making male content too without problem and simple prompts would do just fine. Can't wait for you making more and more futa related checkpoints with different styles
@razzler you can merge lycoris and checkpoints by using the gui implementation of the SD training scripts or the command line. I'm using the bmaltais gui and it can be found in the utilities tab > lora > merge lycoris
https://github.com/bmaltais/kohya_ss
https://github.com/derrian-distro/LoRA_Easy_Training_Scripts
@ckmlai Grazie!!, will def try this