Sign In

Comic Babes

5.2k
34.7k
502
Verified:
SafeTensor
Type
Checkpoint Merge
Stats
7,927
Reviews
Published
Jan 3, 2024
Base Model
SD 1.5
Hash
AutoV2
906F4A33C9

January 2024 Update:

Comic Babes v2 is here!

A new version featuring merges from Alexds9's Babes and LostDog's iComix - two terrific models. Plus, Alexds is a terrific member of the SD community. I highly recommend checking out his discord. There are a ton of great creators and resources on there!

This version is finally enough of a change from Comic Babes 1 for me to upload. I think the lines are a bit sharper, shorter prompts produce better images vs CB1, and it recognizes quite a few celebrity faces.

It does have a tendency to bend women at the waist, but I think that's a result of that pose being so prevalent in comic books. Try different pose/camera prompts if you're getting a lot of "45 degree" ladies. It's also a fairly horny model, so women tend to be a bit top-heavy. More on that below.

By default the model will produce 3d, digital painting style images. For a more cartoon/comic look, I recommend (not all at once, try these one or two at a time):

Positive prompts to try for less-realistic style: cartoonish style, low contrast, flat color style, restricted color palette, line art style, sketch style, (glossy:.7), etc.

If your ladies are too busty, try adding "small breasts, medium size breasts" to your positive prompt, along with "huge breasts" or "large breasts" in the negative. Trust me, they'll still be a solid c-d cup lol.

For my negative prompts, I almost always use the easynegative embedding, along with a few "must have" prompts. CB2 does very well with minimal negative prompts:

Negative prompt: easynegative, 3d, render, photorealistic, huge breasts

Really that's it for the negative prompt. If you're seeing things you don't like in your images, try adding one or two negatives at a time.

Also, Clip Skip 1 and 2 are a lot of fun to mess with. I like both enough, so I can't recommend one over the other.

Inpainting works for me when I'm correcting minor things (eyes, hands, mouth, etc). I haven't tried totally inpainting another image. Your mileage may vary.

Here's my typical workflow:

Model: Comic Babes v2

VAE: whatever you want. I've been using Anything

Clip Skip: Try 1, 2, or even 3. I usually start with 1, but CS 2 listens to style prompts a bit better I think. For this example I used Clip Skip 1

Positive Prompt:

(cartoonish style), flat color style, illustration of a woman, walking down the street, long platinum blonde hair blowing in the wind, wearing winter coat and a scarf, looking happy, looking up at the sky, snowing, snow, cold winter day, busy city background, cars, buildings, detailed face, perfect eyes

Negative prompt:

easynegative, 3d, render, photorealistic

Steps: 21

Size: 512x832

CFG: 7

Batch count: 4

Try this a few times, tweaking prompts one or two words at a time, until I find one I like.

Note: faces might not have much detail at 512x! That's what the next step is for.

Send an image I like to img2img

Just resize

Steps: 30

512x832

CFG: 7

Denoise: .3-.4

Script: SD Upscale

Upscaler: ESRGAN_4x

Tile: 64

Scale Factor: 2

Then inpaint any minor details.

Done.

You can find this image in my examples!

I have a slow machine. I'm actually running on a mac. My average iteration takes 2.5s/it, so I work within the limits of what I have. I'm sure you can produce more and larger images with a better machine.


April 2023 Update:

First of all, thank you so much for 1,000 likes (and counting)! I had no idea this model would do so well. I just found myself using it all the time and decided to share it. It makes me happy that so many people enjoy it! And, I hope this encourages you to share your ai assets with the community as well!

Also, I added version 1.0 Baked Vae. It's the same as the original version, just includes the anything v4 vae. I was getting consistent results with this baked version, and decided to share it with the community! Can't wait to see your images!

A model I merged that I just keep coming back to, so I decided to share it.

A 50/50 blend of iCoMix v1 and Babes v1.1. Recommended you use a vae. I use anything_v4, which is available for download here.

EasyNegative also works very well with this merge, and minimal negative prompts tend to produce nice results.

I'm already working on some other versions that incorporate new versions of iCoMix and Babes.

Please share your images! I'd love to see what you create.

Thanks to lostdog and alexds9 for their terrific models.