Plowcam POV by IsnAI

I can guide you with the best practices, tips, and tricks to achieve your goal in your Stable Diffusion training. Take a look at my Patreon, subscriptions for just $5. Exclusive content. Patreon. :)

About:

Sorry for the absence, but this LoRA son of a bitch was getting on my nerves. 🙄

It's purpose is to generate images in the 'plowcam' style.

It was trained in 768x768, using captions and 143 high-resolution images.

I'm releasing this alpha version because it's the best I could do atm. I'm going to increase the dataset to try and reduce the bizarre images it generates every now and then.

Oddly enough, and I have no idea why, it works better with anime models.

Btw, this model is stubborn. It can generate several good images, but sometimes it starts to produce just crap. 🤬

It's also very sensitive to positive and negative prompts, the slightest of tweaks make big changes in the inference.

This time I didn't even have the patience to choose better images, several have issues with the hands, and some slight anatomical problems, but everything can be easily fixed in 2 minutes with Photoshop or inpaint.

DO NOT FORGET THE DASHES IN PROMPT if you want to replicate the images I did.

Special thanks to stapfschuh and his articles with tips for prompts.

The demo images aren't edited, but they are cherry-picked and generated in A1111 WebUI, using hires-fix with 0.3 ~ 0.5 denoising and 0.1 extra noise multiplier.

Suggested Settings for Inference:

Model: realDream_11

Positive: woman,man,sex-from-behind,frowning <lora:sexFromBehind_by_IsnAI:1:0.8)

Negative: (bad-hands-5, asymmetry, asymmetrical-shapes, sweets, snap-fingers, fat-tits, HD-logo, bodily-deformities, 3D-rendering)

Sampler: DPM++ 2M Karras

Steps: 28

CFG: best results with 7

Resolution: best results in 768x768

LoRA Weights: 1:0.8

Note¹: As you can see, the LoRA tag has two weights: <lora:Plowcam_alpha_sd15_by_IsnAI:1:0.8)
This is a feature of the A1111 (which many people aren't aware of), but you can set separate weights for the UNet and the Text Encoder. The TE is always a pain in the ass to get the training right, so it's usually better to reduce its weight in the inferences rather than having it undertrained.

Note²: You will see big variations in style and images depending on the base model used and minor adjustments in resolution.

Note³: If your images are getting weird and bizarre, you probably didn't reduce the TE weight. If you aren't using A1111 WebUI, you can do the common weight reduction: <lora:sexFromBehind_by_IsnAI:0.8), but you will need to test which weight will suit better.

Triggers:

Main Tags Used on Training: woman, man, sex from behind

Secondary Tags Used on Training: open mouth, looking at camera, closed eyes, penis, busty, frowning, half-closed eyes, smile, looking back, wide eyes, glasses, grabbing her hair