Skip to content

MakeAnything

Generate step-by-step creation sequences across domains

Visit official site

What MakeAnything is

The model is described as a diffusion transformer that generates high-quality step-by-step creation sequences across several domains.

What you can do with it

  • Generate sequences for painting, sculpture or icons
  • Start a sequence from text or from one image
  • Adapt the model to a new domain with few samples

Getting started

  • Clone the repository and enter the project folder
  • Create a Python environment and install dependencies
  • Choose an architecture, download weights and prepare data

Pros & cons

✓ What we like

  • Few samples are enough to adapt to a new domain
  • Pretrained weights and training scripts are provided

! What to watch out for

  • Sequence generation follows the chosen architecture
  • Local training needs an organised dataset and config file

FAQ

Which architectures are listed?

Asymmetric LoRA for text-to-sequence and Recraft Model for image-to-sequence are named.

What resolutions are stated?

The features name 1024x1024 and 1056x1056 sequence generation.

Last reviewed: 2026-11-15

More AI image generation tools

View all →

How we review