Sign in Submit site

deep-floyd/IF

github.com/deep-floyd/IF AI Image Generators Open-source project

DeepFloyd IF is an image generation model project hosted in this GitHub repository.

Visit deep-floyd/IF 24 alternatives
github.com/deep-floyd/IF
deep-floyd/IF homepage screenshot

About deep-floyd/IF

DeepFloyd IF is a text-to-image model project from DeepFloyd Lab at StabilityAI, hosted in this GitHub repository. It uses a frozen T5 text encoder and three cascaded pixel diffusion modules. A base model generates 64x64 pixel images from a text prompt, and two super-resolution models increase the output to 256x256 and 1024x1024 pixels. The README describes the model as open-source and reports a zero-shot FID score of 6.66 on the COCO dataset.

Running the model requires substantial GPU memory. The README lists 16GB of vRAM for the base and first upscaler stages, and 24GB when the 1024x1024 upscaler is included. Using the model also requires accepting its usage conditions on Hugging Face and logging in with an access token. The repository includes Jupyter notebooks for local use and integration with the Hugging Face Diffusers library.

Key features

  • Three cascaded stages producing 64x64, 256x256 and 1024x1024 pixel images
  • Frozen T5 text encoder for interpreting text prompts
  • Local Jupyter notebooks for Dream, Style Transfer, Super Resolution and Inpainting modes
  • Integration with Hugging Face Diffusers for running each stage separately
  • Inspection of intermediate results when using the Diffusers integration
  • Model CPU offloading in Diffusers, with a stated minimum of 14 GB vRAM
Best for: Python developers and researchers with a suitable GPU who want to run a text-to-image model locally and customize its stages.
Good for
  • Modular design lets users customize each generation stage and inspect intermediate results through Diffusers
  • Provides notebooks for several image modes, including style transfer, super resolution and inpainting
  • The README reports a zero-shot FID score of 6.66 on COCO, a figure the authors state themselves
Watch out for
  • Requires at least 16GB vRAM for the base and first upscaler stages, and 24GB with the 1024x1024 upscaler
  • Users must accept the usage conditions on Hugging Face and log in with an access token before use
  • Older torch versions require xformers and the FORCE_MEM_EFFICIENT_ATTN=1 setting

deep-floyd/IF at a glance

Type
Open-source project
Pricing
Not confirmed. Check the site
Open source
Not stated
Language
English
Last checked
Oct 6, 2026 · Online when checked
Listed
Oct 5, 2026
python

deep-floyd/IF alternatives

Tools and sites you could use instead, ranked by how closely they match.

All 24 alternatives

Popular in AI Image Generators

Well-known listings from the same topic.

All 363
FigureLabsfigurelabs.aiImagePromptGuruimagepromptguru.netAI SVGaisvg.artCGDreamcgdream.aiFlux.1 AIflux1.aiPixraypixray.com
Related topicsAI Business ToolsAI Infrastructure & MLOpsAI Agents & AutomationAI Data & AnalyticsAI Chatbots & AssistantsAI Voice & AudioAI Writing ToolsAI Video Generators

deep-floyd/IF FAQ

What GPU memory is needed to run DeepFloyd IF?
The README lists 16GB of vRAM for IF-I-XL and IF-II-L, and 24GB when the Stable x4 upscaler to 1024x1024 is also used. Diffusers with model CPU offloading is described as running the pipeline with as little as 14 GB of vRAM.
What steps are needed before using the model through Hugging Face?
Users need a Hugging Face account, must accept the license on the DeepFloyd/IF-I-XL-v1.0 model card, and must log in locally with a Hugging Face access token.
Is something wrong or out of date?

Tell us and an editor will check it.

Own this website?

Claim the listing and suggest updates yourself. An editor checks every claim first.

Claim this listing

Listing maintained by Chris McClure and the WebIndexly editors · Updated Oct 13, 2026 · How we rank