Get a weekly rundown of the latest AI models and research... subscribe!



AI model preview image
The Activating More Pixels in Image Super-Resolution Transformer (AMPIST) is a model that aims to improve the quality of super-resolution images. It does this by effectively leveraging the self-attention mechanism in transformer networks, which helps to capture dependencies between pixels in an image. The model incorporates a pixel-wise feature transformer that generates attention maps for each pixel, allowing for more accurate and context-aware information aggregation. Experimental results show that AMPIST outperforms previous state-of-the-art models in terms of improving the perceptual quality and sharpness of super-resolution images.

Use cases

The AMPIST model has several potential use cases in the field of computer vision and image processing. One possible application is in the enhancement of low-resolution images, particularly in fields such as medical imaging or surveillance where visual clarity is crucial. By utilizing the self-attention mechanism and pixel-wise feature transformer, AMPIST can generate high-quality super-resolution images with improved perceptual quality and sharpness. Another potential use case is in the restoration of degraded or damaged images, such as old photographs or historical documents. The model's ability to capture dependencies between pixels and generate context-aware attention maps can help to restore missing or damaged details, resulting in clearer and more accurate representations. Additionally, the improved image quality provided by AMPIST opens up possibilities in areas such as digital art, where artists can create high-resolution artwork from low-resolution sketches or images. Overall, the AMPIST model has the potential to be incorporated into various products and practical solutions, ranging from image editing software to medical imaging technologies.


Cost per run
Avg run time
Nvidia T4 GPU

Creator Models

No other models by this creator

Similar Models

No similar models found

Try it!

You can use this area to play around with demo applications that incorporate the Hat model. These demos are maintained and hosted externally by third-party creators. If you see an error, message me on Twitter.

Currently, there are no demos available for this model.


Summary of this model and related resources.

Model NameHat
Activating More Pixels in Image Super-Resolution Transformer
Model LinkView on Replicate
API SpecView on Replicate
Github LinkView on Github
Paper LinkView on Arxiv


How popular is this model, by number of runs? How popular is the creator, by the sum of all their runs?

Model Rank
Creator Rank


How much does it cost to run this model? How long, on average, does it take to complete a run?

Cost per Run$0.01045
Prediction HardwareNvidia T4 GPU
Average Completion Time19 seconds