RealCartoon3D

ANIME FEMALE DIGITAL ART MECHA MIX SCIFI FANTASY ART ART STYLE ILLUSTRATION BEAUTIFUL CARTOON FANTASY GIRLS MAN REALISTIC VARIETY LANDSCAPE 2.5D SEMI-REALISTIC AFRICAN-AMERICAN STYLES REAL CARTOON 3D

Want to buy the author some coffee? (Buy a cup) - https://ko-fi.com/7whitefire7

• Pruned Versions are under the files section (If you do not have it set to select Pruned already)

Please click "read more" to see more detail on the "why" for each version of this model.

**07/15/23: Added Version 4. I hope you all enjoy! Thank you all for the reviews and likes! Appreciate the support!

**07/16/23: Had to update cover photos because they were too "mature" for stats to count. :/

...Please Share your images...Love seeing the results...

This is my first checkpoint upload. I have been testing out different models (DreamShaper and Duchaitenniji being my favorites) and loras and VAEs to get a look that I really like. This checkpoint merger is a mix of many (I found others with a style I liked, and used them just to influence the overall look). So the weights varied, and I would re-merge to the newly merged checkpoint to get more subtle changes. I was going for a 2.5, realistic, cartoon, anime look. This model does well on a range of LoRAs I have put on it. Along with pretty much being able to do what I ask it to do. I also did not want all the characters to all look asian. I tested it with some various looks and it did alright. I hope the samples give a good idea of the look and abilities.

VAE is Baked in. ALL THE SAMPLES WERE RUN WITHOUT A VAE. MOST ON FULL VERSIONS.

The normal settings I ran it at are as follows:

• Width: 512 (Have done up to 568)

• Height: 768 (Have done up to 900)

• Sampling Method: DPM++ SDE Karras (Also DPM++ 2M Karras and Euler A do well)

• Sampling steps: 27-30 normally (27 is a nice spot for Euler A)

• Hires.fix settings: Upscaler (R-ESRGAN 4x+ and 4k-UltraSharp), Hires Steps (10), Denoising Str (0.34 - 0.45 normally), Upscale (2)

• Clip Skip: 2

• Ran ADetailer to help enhance the eyes but mainly if I was doing a DPM sampler

About Version 1.0:

First attempt at this. Good base, and does most of what I wanted. Loved the look. I knew I would have other version up; but thought this one was at a good spot to share.

About Version 1.5:

Have done some more subtle updates. Version one is still very nice; but wanting to see what I can do with it as a base now. I also want the checkpoint to do well when you ask for other ethnic types and just general requests period. This version has some samples that are similar to show difference from version one, and a couple to show some of the ability. The two beasts scene has no LoRA.

....All the settings are pretty much the same for getting images from Version one...

Negative prompts I do use often are: BadHandV5, Deep Negative, BadHandV4,

...Would love your feed back on both versions. Thank you for downloading and enjoy!!...