Instructions to use ideogram-ai/ideogram-4-nf4-diffusers with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use ideogram-ai/ideogram-4-nf4-diffusers with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("ideogram-ai/ideogram-4-nf4-diffusers", dtype=torch.bfloat16, device_map="cuda") prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k" image = pipe(prompt).images[0] - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- Draw Things
- DiffusionBee
A question
Hi Ideogram team, thanks for releasing Ideogram 4 open weights.
I have a question / feature request about local image editing support.
On the Ideogram web app, uploaded images can be used with features such as Remix, Magic Fill, Open in Studio, and image reference workflows. However, the current open-weight release on Hugging Face (ideogram-4-fp8, ideogram-4-nf4, and ideogram-4-nf4-diffusers) appears to expose only text-to-image inference locally.
Is there any plan to release local image-to-image / image editing support for Ideogram 4, such as:
- using an input image as a reference during generation
- remixing an existing image while preserving composition and object geometry
- inpainting / Magic Fill with masks
- changing scene attributes, for example turning an existing drone photo into a rainy-day version while preserving the drone structure
If this is already possible with the current open weights, could you provide an example script or documentation showing how to pass an input image and optional mask to the local pipeline?
If it is not currently supported, it would be very helpful to know whether the web app’s Remix / Magic Fill functionality depends on additional models or conditioning modules that are not included in the current Hugging Face release.
Thanks again for the open release.