Log in

Create custom AI model

Current avg. waiting time is 1h 38m.


Model type

Purpose of the model


Transform yourself into anything

Product shot

Capture objects in many scenerios


Replicate custom art style

be as general as possible, do not enter names or proper nouns here

Model style

Base model used for training

ICBINP Afterburn

Realistic Vision v3

Realistic Vision v5.1

Openjourney v4




Dark Sushi Mix v2.25


Stable Diffusion 1.5

Stable Diffusion 2.1

Realistic Vision v1.3


Openjourney v1


Upload images

Set of training images

Upload ~20 images of the same person/animal. Make sure that:

  • There are no duplicated images
  • Uploaded images are widely diverse
  • Images are larger than 512x512px

It's also best to:

  • Use photos with different backgrounds and angles
  • Have no other people in your photos
  • Use photos with good lighting
  • Use photos with different framing (close ups, side, chest up, full body)
  • Use photos with different facial expressions
  • Have photos where you look into the camera and other photos that you look away the camera
  • If couple, make sure that every photo has both people in it
  • Use photos without sunglasses, hats, excessive makeup etc.

For the most diverse results use:

  • 10 close up photos
  • 5 upper body photos
  • 3 full body photos
  • 2 side profiles photos

Please also note that the training process is performed on images with 1:1 ratio, so:

  • Every non-square picture will be cropped at the center
  • Make sure that your subject is fully present in the cropped pictures below

Do not:

  • Create model on photos of children or minors.

The better you follow these instructions, the better your results will be!


Advanced settings

Training parameters for pros

Change these only if you know what you're doing!

Modifier used in prompt to invoke the new style. Needs to be unique.
What is this model of?
Total number of training steps to perform (10k max).
Minimal class images for prior preservation loss (10k max).
Initial learning rate (after the potential warmup period) to use.
Number of steps for the warmup in the lr scheduler.
The scheduler type to use

Model details

Details used to identify your model

Used to identify your model in tools. It does not affect training.

Train your model

Start the process

Before submitting, please note that:

  • You'll get model hosted on ready to use at any time and files to use the model on your own machine.
  • Quality of the model mostly depends on quality of used images.
  • We do not offer refunds, as we start incurring costs immediately.
  • We delete uploaded images as soon as training is finished.
  • You will get some pre-generated images to start (we're still working on our prompt database).
  • Current avg. waiting time is 1h 38m.
  • It costs 2000 credits to create a model.
  • With your current plan (), you can have 0 models stored.