Pterosoft Studio has released Clarity 1.1, a major optimization update for its free local AI image upscaler. The new version reduces the application’s size, lowers memory consumption, shortens initialization and introduces a redesigned interface.
Clarity 1.1 also resolves a performance problem affecting its dedicated texture model. This fix makes the application better suited to game-development textures, materials and other detailed graphical assets.
Clarity Now Uses the Advanced DAT Architecture Exclusively
The most significant change concerns Clarity’s underlying AI architecture. Previous versions included both SwinIR and DAT models.
Clarity 1.1 removes the now-obsolete SwinIR architecture. The application now relies exclusively on the more advanced DAT architecture for AI upscaling.
Maintaining two separate architectures increased the application’s size and resource requirements. It also added components that no longer contributed to Clarity’s future development direction.
Consolidating the application around DAT allowed Pterosoft Studio to remove hundreds of megabytes of unnecessary files. It also simplified Clarity’s internal AI pipeline.
Application Size Falls from 818 MB to 138 MB
Removing SwinIR produced an immediate reduction in the application’s overall size.
Clarity 1.0 required approximately 818 MB of storage. Clarity 1.1 reduces that figure to only 138 MB.
That represents a reduction of approximately 83 percent. The new version is around one-sixth the size of the previous release.
The smaller package requires less storage space and makes Clarity considerably faster to download. It also simplifies distribution and future updates.
This reduction does not come from removing Clarity’s primary AI-upscaling capabilities. Instead, it reflects the decision to concentrate development on the DAT architecture.
RAM Consumption Has Been Cut in Half
Clarity 1.1 also significantly reduces memory consumption.
The previous release used approximately 8 GB of system memory during AI processing. The new version lowers that requirement to approximately 4 GB.
Cutting RAM usage in half makes Clarity more accessible to users with mainstream hardware. It also leaves more memory available for other applications running alongside the upscaler.
This improvement can be particularly useful for game developers and digital artists. Their workflows frequently involve memory-intensive applications such as game engines, image editors and material-authoring tools.
Actual memory usage can still vary according to the selected model, source image and output resolution. However, the architectural changes substantially reduce Clarity’s general memory footprint.
Faster Application Initialization
The removal of the unused SwinIR components also improves initialization times.
Clarity now has fewer AI components to prepare when starting a processing session. As a result, users can begin upscaling images more quickly.
This improvement is especially noticeable when repeatedly opening the application for smaller tasks. It also makes Clarity feel more responsive during everyday use.
Texture Model Performance Has Been Fixed
Version 1.1 resolves a performance issue affecting Clarity’s dedicated texture model.
The problem prevented the model from delivering its intended processing performance. The update corrects that behaviour and restores the model’s expected speed.
Clarity’s texture model targets assets such as surface textures, scanned materials and game-development resources. These images often contain repeated patterns, small surface variations and fine structural details.
The fix makes the model more practical for artists who need to upscale multiple textures or process high-resolution production assets.
A Completely Redesigned Interface
Clarity 1.1 also introduces a new user-interface design.
The redesigned interface gives the application a cleaner and more modern appearance. It presents Clarity’s upscaling and resizing tools through a more polished visual experience.
Users can still access Clarity’s core AI-upscaling features and traditional resizing methods. These include point sampling, bilinear interpolation, bicubic interpolation and cubic B-spline interpolation.
The interface update changes how Clarity looks and feels without compromising its focus on a simple local workflow.
What Has Changed in Clarity 1.1?
The main improvements include:
- Removal of the obsolete SwinIR architecture
- Exclusive use of the advanced DAT architecture
- Application size reduced from 818 MB to 138 MB
- RAM consumption reduced from approximately 8 GB to 4 GB
- Faster initialization
- A redesigned user interface
- Fixed performance for the dedicated texture model
Free AI Upscaling That Runs Locally
Clarity remains completely free and processes images locally on the user’s Windows PC.
Images do not need to be uploaded to a remote server. This gives users greater privacy and removes cloud-processing queues, usage credits and subscription requirements.
The application uses DirectML GPU acceleration to make use of compatible graphics hardware. Clarity supports photographs, text, digital artwork and game-development textures.
Version 1.1 represents an important step toward making the application smaller, faster and more accessible. By concentrating on DAT, Pterosoft Studio can focus future development on a single advanced AI architecture.
Clarity 1.1 is available now as a free download.