Hello, we are excited to announce a very significant update to Gigapixel with version 7.1.0 today. You can dig more into the details and changelog notes on this release thread, but we’ll cover some of the major changes in this post, as well as what’s coming next.
See the release thread here. Direct download links are available in that thread, but you can also use the in-app updater.
This release marks an important milestone for Gigapixel. In short, a grand total of 3 new incredible AI upscaling models have been added into Gigapixel 7.1.0, comparison view returns, we have implemented a new standardized Topaz UI framework, and to wrap it up we also added face selection, and crop rotation.
Before we launch into the update, we also want to talk a little bit about the previous release of Gigapixel version 7.0.0, and why we plan to continue updating and building out new features for Gigapixel alongside Photo AI and Video AI.
Looking Back on a Super Sized 7.0.0
Out of all our three Image Quality products, one application made a return to the limelight releasing its seventh version earlier this year.
Gigapixel AI ushered in a refreshed user interface with simplified tool tips for more preview visibility, as well as realigned our Preferences and Export Settings for better workflow.
Above: A side by side comparison of Gigapixel’s User Interface (Left: v6.3.3/Right: v7.0.0)
The upscaling application also made its comeback with revitalized AI models. The Standard v2 edition saw a reduction in blurry patches. High Fidelity v2 preserves grain consistency and gives a more natural textured look, especially for photos taken with modern large-sensor cameras.
Standard v2 improves blurry patches and details compared to the Gen1 Standard model
High Fidelity v2 maintains more naturalized textures than before in this example
Because we know our upscaling was what put Gigapixel AI back on the map, it was imperative to implement a new code base to support the power behind these AI models as we scale beyond a billion pixels.
Our team has since then spent the last few patch updates monitoring bugs, re-establishing a fluid user experience, and optimizing performance leading up to this next release.
More recently, we’ve drastically sped up image loading with a personal benchmark of 20,000 images loaded in seconds and CPU usage reduced within milliseconds after compressing (Windows only, Input <1000px). Significant results we are proud to share.
We are excited to continue to expand what is possible in Gigapixel. Without further ado, we’d like to share with you what we’ve been building for the release of Gigapixel 7.1.0
Gigapixel 7.1.0 Arrives
Version 7.1.0 Updated User Interface
New “Recovery” AI Model for Upscaling (Beta)
A new player joins the Gigapixel AI Model line up that significantly improves quality for low resolution input images by intelligently generating minute details. While the Recovery model is slower than the other upscaling models, the results are truly stunning, and can be worth the wait.
Left: Original Input 610x350px. Right: Upscaled at 6x (3660px x 2100px)
Recovery is a diffusion-based generative AI upscaling model, which produces different results than our other upscaling models. The model performs very well when upscaling low-quality source images , for example images with low resolution or significant blur.
The model also seems to perform well on faces, buildings/structures and other objects. The detail slider allows you to control the level of generated detail in the image after the initial render in post-processing. We have found that detail settings in the 50-75 range work well for most scenarios.
Many of our beta testing users have generated incredible image recoveries on otherwise unrecoverable images such as old family photos for example. We are really excited to see what you can do with this new model.
We have plans to release significantly more performant and smaller versions over time – but you will need a powerful graphics card for this version of the model. We recommend upscaling images under 1000px by 1000px depending on the size of your graphics card. This model currently requires 6GB+ of VRAM, and runs best at 8GB of VRAM and above. Renders with the Recovery model may take several minutes even on relatively small images at the moment.
IMPORTANT: Even with large GPUs, upscaling images larger than 1000px may affect your system hardware due to lengthy runtimes, and is not recommended without careful monitoring. This model is also optimized to generate the best results on smaller images.
New “Text & Shapes” Model
Before/ After of improved text using Text & Shapes AI Model
We have added a new AI-upscaling model that enables more distinct patterns and shapes during upscaling with man-made objects or defined shapes such as cars or buildings, and even written text and fonts in multiple languages. The results we’ve been getting for these types of images are better than ever before with any of our prior models.
New “Low Resolution v2” Model (Beta)
Low Resolution v2 preserves more realistic details without overdoing contrast
The third model coming to Gigapixel 7.1.0 is a refresh of the Low Resolution AI upscaling model. This model has been upgraded into the second generation family of models, performing well out of the box for both low resolution and high compression images.
You will still have access to the Gen1 model as well, using the drop-down menu on the top right of the model selection panel.
Comparison View
A Snapshot of Comparison View in Use
The beloved comparison view makes its return debut with its notable four quad design. We’ve cleaned up how the interface is used, showing exactly which image you’re working on.
You can also now configure default settings for which AI models are used in the view and remember those settings for future use. We will continue to improve and support this view over time, and we have some more exciting updates in store.
Face Selection
Above: Face selection panel is available when Face Recovery is enabled
Face Selection has been added to Gigapixel, which will allow you to see the number of faces detected and manually choose which faces should be recovered. This is useful in scenarios where faces have been detected incorrectly, or when you simply don’t need to recover certain faces in your image.
Crop Rotation
Crop panel now allows for manual crop rotation
We have extended the Crop tool within Gigapixel, adding new capabilities into your workflow. Images can now be rotated to create perfect vertical or horizontal lines.
What’s Next for Gigapixel?
Our focus on the future of Gigapixel will be to bring physically accurate color and scale to your images, extending the existing capabilities to prepare your powerful upscaling results for showcasing in the real world, including for printing.
This involves:
- Increasing color accuracy and our handling of color spaces and color profiles
- Increasing resolution beyond the current limitation of 1 billion pixels
- Improving existing upscaling model capabilities for art and CG assets
- Adding more ways to print both inside and outside of Gigapixel
- Adding more capabilities to cropping at physical sizes
- Adding back the ability to compare model outputs in detail
Beyond print, we are also expanding the ability to share and to display your images from Gigapixel to the world in the highest quality possible.
If you’re interested in getting a sneak peek, please apply to join our Beta Program. Users who join get a little extra perk for participating. We can’t wait to hear your thoughts and to make Gigapixel even better with your help.
Cheers,
Dakota Wixom
Product Manager, Topaz Labs
