<h1 id="penis-size-slider">Penis Size Slider</h1><h3 id="tom-trained-on-moody">TOM = Trained on Moody</h3>The TOM models is models that is trained directly on Moody models instead of stock Z-image. And because of this i will not provide any support, or take any criticism on these models if they dont work propperly on other models or with other LoRAs. They will probably work fine for most other models, but for best results i suggest using them with Moody V10 R1 as this is the only model i have tested the sliders on. After i got the models public i tested some images with the on-site model from PornMaster, and all the v2.0 models made pretty good slongs on about 2-3 images on a 8-image batch, witch is actually very good ratio on a PornMaster model imo.<h3 id="this-lora-does-not-make-penisesit-only-changes-the-size">THIS LORA DOES NOT MAKE PENISES IT ONLY CHANGES THE SIZE</h3><h1 id="v2:">V2:</h1>As you probably see, v2.0 is 3 different models. I could have named them v2.1 or v3.0, but they are not upgrades from eachother. They are made to cater different requests i have gotten on v1.0. V2 does not always make larger penises compared to V1, but it should be much more stable.I suggest you stick to the "main" version because this is the most stable of them. "Thick" and "Long" is special models for very spesific cases where main fall short in girth or length. But be aware that thick and long is very unstable, and can add some extra dicks or balls at higher values. And they also affect the persons in the image to a high degree, at least compared to my usually extremely high standard on other Sliders i have. All v2.0 models are normalized to be 0-10 range sliders. But 10.0 is usually to much for a good result. So i will recommend max 8.0<h2 id="recommended-range-all-v2-models:-2.0-to-8.0-(some-times-10.0)">Recommended range all V2 models: -2.0 to 8.0 (some times 10.0) </h2><hr /><h1 id="v1:">V1:</h1>I understand why noone has made this before now. This felt almost impossible to make, for multiple reasons. But i have managed to make a pretty good and somewhat stable LoRA. You can see from the samples, that all use 2 different seeds, since there was no need to "cherry pick" sample images from multiple seeds. It should work with most, if not all checkpoints and LoRAs that make penises to begin with. There is some things to keep in mind, that i dont think is possible to fix. Z-Image have a extremely strong relation to the persons height and penis length. So with longer penises come taller persons, both male and female in the scene. And the height affect a whole new pool of problems, like makeup etc. So this is not the cleanest LoRA i have made, by a long shot, but it is at least a V1 that works pretty well.<h3 id="a-few-tips-and-things-to-keep-in-mind:">A few tips and things to keep in mind:</h3><ul><li>If there is no space for a larger penis in the image, it will zoom out, instead of trying to force the penis to fit.</li><li>Not the best at 1:1 aspect ratio. very wide, or very tall seem to make the most room for it to grow and therefor also the best results.</li><li>Solo male/futa where nothing is in front of the penis tend to be extra sensetive.</li><li>Not very good at negative, but it works.</li><li>Prompting always helps a lot, so a combination of this LoRA and prompting is best.</li></ul>I have not been able to test it with all checkpoints and LoRAs, but the ones i have tested works from very good (Moody V10) to ok-ish (BigLove 3). I would highly recommend Moody V10 instead of V11 or V12.<h3 id="recommended-ranges-v1:-general:-2.0-to-5.0moody-proporn-v10:-(most-stable-checkpoint-i-found)-5.0-to-8.0-(some-times-up-to-10.0)mystic-xxx-v5:-2.0-to-5.0(most-stable-lora-i-found.-but-i-have-not-tested-all-mytixx-loras)">Recommended ranges V1: General: -2.0 to 5.0 Moody Pro/Porn V10: (Most stable checkpoint i found) -5.0 to 8.0 (some times up to 10.0) Mystic XXX V5: -2.0 to 5.0 (Most stable LoRA i found. But i have not tested all Mytixx LoRAs)</h3>

Penis Size Slider - (ZIT)

TOM = Trained on Moody. This is the main v2.0 model. Its the most stable of the v2.0 variants, and should be the pref one to use in most cases. This LoRA was trained directly on a Moody Base model, so use it with Moody for best results. But it also works ok on other models.

Files with SHA256: 9d537bd9b8f44463aebff28ae64f65301c93c7cdb411121411cbf46d5f61f07e