This is a simple character LoRA. Meet Cameron. You can call him "Cam" or "Cam-o" because he's totally chill like that. Cam is a college student who enjoys staying in shape and socializing with his peers. He should be fairly versatile, for use with other LoRA's and checkpoints, that is.

Pro Tip

Character models can be a great resource for perfecting your generated images when using image-to-image and image-to-image-inpainting. Have characters in your image foreground or background that are blurry and or under-defined? Use a character LoRA to help SD enhance your image. Image-to-image-inpaint is the most effective use of this strategy. Cover your subject with the masking brush and choose the settings for inpaint mask only. Then set your denoising strength from .10 - .68, depending on how much change or definition you would like.

V2

This version is not a variation of V1, it is a total overhaul. I retrained the model using a refined data set and better training parameters. The results are a much more consistent look to the character and higher quality images. All the training images I used were my original creations, as is consistent with the training of all the LoRAS I have published.

Cameron has distinctive two-tone brown eyes. His hair is designed to be short to medium length of either a blonde, brown or black color.

The following prompts can help you guide his generation. The trigger word is cameron.

Positive Prompts

blonde_hair - black_hair - brown_hair - short_hair - very_short_gair - buzz_cut - spiked_hair - undercut - fade_(haircut) - toned_male - bara - yaoi - hoodie - jeans - t-shirt - print_shirt - tank_top - shorts - boxer_briefs - sneakers - pool - male_swimwear - swim_briefs - swim_trunks - bedroom - outdoors

The following negative prompts are only suggestions and aren't specific to this model, just prompts I find useful when needing to refine my generations -

Negative Prompts

1girl - pussy - simplified - abstract - unrealistic - impressionistic - low resolution - blurry - low quality - bad anatomy - lowres - worstquality - bad propportions - out of focus

*About prompt syntax

The use of underscores, such as in the prompt blonde_hair, is for triggering danbooru tags. Additionally, underscores can be used between words to help Stable Diffusion focus on considering the connected words as a whole, decreasing the chance that the Stable Diffusion will process two tokens seperately that a person using natural language might instinctively know are connected to mean a specific thing.

Some tokens Stable Diffusion may understand as being a single token even without the underscore, such as bad anatomy. Some prompts seem to be understood as a token where the words are written together without a space or underscore beween them, this seems to be true when prompting colors, such as darkred or mediumturquoise.

I can't say with 100% certainty my thoughts on prompt syntax are correct. My experience seems to confirm that my explanation is correct. When I list prompts in my model descriptions I try to be precise about how they are written. I write them the way that Stable Diffusion understands them, to the best of my knowledge, they aren't typos.

V1

The Checkpoint I used to create the demonstration images was Starstruck and I have linked it in the 'suggested resources' section below, though I verified the model as compatible with a variety of Checkpoints available here @ CivitAI. I will say that the checkpoint used with this LoRA has an impact as to which timeline your Cameron hails from.

DPM++ 2M at 38 steps works well as does

DPM 3M SDE at 40-90 steps.

This model was trained to be anatomically correct. Using this model with the Starstruck Checkpoint will draw reliably anatomically correct images a majority of the time. Stable Diffusion has a strong bias towards drawing females. The tags "1girl" and "pussy" placed in the NEGATIVE prompt box can be very helpful at neutralizing this bias. The tag "male_focus" in the main prompt also helps.

Cameron is a college aged lad with brown_hair and brown_eyes. Prompting those characteristics will tend to produce an image with maximum Cam-ness, but he can have blonde_hair or red_hair, ect, if you wish. It's just like whatever for him, he doesn't mind. He misses his girlfriend, she goes to an out-of-state University, 🙄 but no worries, more time to hang with the bros.