For RAW images, instead of estimating the missing color information for each pixel based on the values of neighboring pixels, the AI model uses context from the entire image. It also considers and compensates for noise that hinders traditional 

The RAW models are activated when the program detects a file with 

 RAW Denoise is developed for the aggressive RAW noise that Bayer sensors have, (Canon, Nikon cameras for example) whereas Fujifilm cameras have a X Trans sensor, and will not have this overall aggressive RAW noise. This is why RAW Denoise activates when it detects a Bayer sensor camera. For any noise on these Fujifilm files, it is recommended to use the Denoise enhancement.

 tab to select the AI model and adjust the settings. 

 model to treat the vast majority of RAW images since most cameras have built-in noise reduction.

This model is reserved for severe noise due to very high ISO (above 6400 ISO). Use it to recover details in images shot in low light.

Open the Photo AI Main Menu > Preferences > RAW denoise

 - select from the dropdown menu to choose between AUTO, RAW Normal, or RAW Strong as your go-to setting

 - adjust the model strength slider to a default setting

 - use this toggle to correct hot pixel issues. A hot pixel is a pixel that doesn't react to light or reacts in an unusual way, appearing as a bright dot in an image. Hot pixels can be caused by sensor age, contamination during sensor production, long exposure times, high gain settings, high sensor operating temperature, or manufacturing defects.

RAW Denoise

This is the first AI model of its kind, built specifically to handle one of the most time-consuming tasks in photo restoration—removing dust, scratches, and surface damage.

Dust and Scratch is a new enhancement released in Photo AI v4.0.0. To know what version your license gives you access to, see your “

The Dust & Scratch AI model has higher requirements for processing than other AI models. 

See the full System Requirements page here. 

8GB VRAM is required for processing locally. 

Mac with Intel CPU/AMD Graphic cards: The stable diffusion technology used on this AI model is 

Mac M-Chip machines (Apple Silicon): Requires 

If your System is under the requirements for this enhancement, you can still process using Cloud Credits. 

For anyone working with scanned film, archives, or damaged photo collections, manual cleanup can take hours or even days to do. There are few, if any, automated solutions on the market today—until now. This two-step workflow combines a smart Dust & Scratch model with a fast and intuitive Healing Brush, offering a complete, time-saving solution right inside Photo AI.

Warning: Batch Processing is not possible with Dust and Scratch.

—it’s designed to give great results right out of the box and should be used 

. The healing tool gives you added flexibility to fix trouble spots quickly and naturally after the initial render is finished.

NOTE: The first usage will be slow as the model needs to load.

" for quicker processing or if your machine does not meet the system requirements for this model or "Process on 

Processing on the cloud will be faster, but will require 

. Processing locally will use your computer's graphic memory. Make sure that your machine meets the 

Once the Dust & Scratch removal is complete, you’ll move on to Healing. Use the 

 with larger scratches or damage. As soon as you let go of the mouse or pen, the app will 

Once you’ve cleaned up the damaged areas, simply click 

It's possible to reopen the Healing brush tool by clicking on it in the enhancement stack to do more fine-tuning if necessary.

Our Phd researchers are working incredibly hard to improve this model EVEN MORE, but sometimes your images may be slightly too soft, over-processed, or details from the background may be removed.


Feel free to send us your original file, your current results, and any feedback to support@topazlabs.com (mailto:support@topazlabs.com). We're here to help you achieve the best results and share your input with our development team to continue improving the tool.

If your file has text in some areas - you can resurrect some of these details using the 

Super Focus is also excellent when combined with Dust & Scratch for old photos, which are also soft. Check out this example:

Dust & Scratch

 is a generative AI tool available in Photo AI that has the ability to fix blur and extract detail. This tool acts as a alternative solution for those stubborn and more difficult images that cannot be enhanced through Photo AI’s other sharpening models.

Batch Processing is not possible with Super Focus when rendering locally or on the cloud.

Super Focus is good to help recuperate files that the 

 filter cannot recover. It is recommended to try the Sharpen filter first, as its much faster and can be implemented in batch processing, whereas Super Focus cannot. 

The Super Focus AI models (v1 and v2) have higher requirements for processing than other AI models. 

Mac with Intel CPU/AMD Graphic cards: The stable diffusion technology used on these AI model is 

If your System is under the requrirements for this enhancement, you can still process using Cloud Credits. 

To use Super Focus, open the image and then: 

If you are on Photo AI v4.0.0 and up: click on the 

 If Autopilot is turned off, the button will be called 

If you are on a Photo AI v3, click on the blue 

See details on the interfaces for each version here

 - see #4 for the Sidebar Navigation Panel information. 

Select the Super Focus version you'd like to use and follow along with the Super focus interface to make your enhancement.

Originally released in v3.3.0, Super Focus V1 introduced a better method to sharpening images unlike it's Sharpen models. 

When using this version, results best shown on images that are at least 4 megapixels or larger images. For best results, recover faces after Super Focus V1, especially on small to medium sized faces as face details are altered by this model. Additionally, use preserve text after Super Focus V1 on small to medium sized text as it may be altered.

Super Focus v2, released in v.3.6.0, is a new model architecture that delivers much faster processing compared to the previous version of the Super Focus model.

Select Super Focus V2 in the dropdown menu

Pick a focus boost depending on your blur level. Extreme blur should use Major focus boost.

Use quick preview to process and review a section of the image

Use selection to display selected areas of the image*

In post processing change the selection or strength to tune your result

Change focus boost to return to the processing menu and reprocess the image

* Important: 
- For any Custom Selection, this will have to be done once the render is made. After Super Focus is processed, you will be able to tweak a custom selection. 
- No matter the selection options made prior to rendering, this AI model will still render the whole image, which will allow you to tweak your selection once the render done, and avoid needing to reprocess. If you have Selection options applied, the preview will only show these affected areas. 

Use Focus Boost only on large images with a lot of blur

For smaller images, avoid focus boost entirely, as the effect may be too strong

If your image has a shallow depth of field use the selection to choose the subject

Use selection to hide artifacts or unwanted detail in the background

Run face recovery and text preserve after Super focus for best results

In some instances, motion blur may be corrected using Super Focus but the tool is not designed to handle extreme cases. Try exploring the Motion blur Sharpen AI model instead.

Super Focus

 uses generative AI to eliminate unwanted objects from your images. By analyzing the surrounding context not only erases the object, but also generates a background that blends with the rest of the image.

Due to the resource-intensive nature of this feature, the Remove tool is currently not supported on Macs with AMD GPUs. It also requires macOS 13 and higher to run with the most optimized results. 

After clicking "Remove object" from the right side panel, a status bar update appears in the bottom left corner of the Preview screen indicating "Locking edits". 

This status ensures that Photo AI locks your enhancements in order to safeguard the generative AI model output and prevents reprocessing generative models which can change the results. 

It also processes all enhancements before the generative tool and prevents further changes to those enhancements.

Locking edits means you cannot make adjustments to other enhancements like sharpen, denoise, or color balance until the "Remove Object" process is completed.

To modify locked enhancements, the generative enhancement must be deleted. You can undo the removal and apply enhancements again after the object has been removed.

 tab to adjust the brush and control the processing. 

The Remove Tool's performance depends on the computer's system profile and the selected editing priorities. Prioritizing speed conserves resources and reduces processing steps on weaker computers while prioritizing quality yields optimal results when hardware performance is not a constraint.

 button to undo the most recent removal process. Click the 

 button once you are satisfied with the mask to remove the object.

Switch between the add and subtract mask buttons to refine the mask if necessary.

Choose between the Regular brush and the Object Selection brush. The Object Selection brush identifies and selects entire objects.

 (such as denoise, sharpening, and color adjustments) before you use the Remove Tool. Make this the last step in your Photo AI workflow to prevent unexpected outcomes where earlier enhancements get locked or nullified during the object removal process.

Keep mask sizes below ¼ of the total image size. Process each object individually when dealing with multiple objects requiring separate masks. Ensure that the maximum distance between masked objects does not exceed 2000 pixels.

Mask the entire object rather than a partial section. Include the shadows and reflections of the objects. If those are not removed, the AI model may attempt to replace the object with something similar.

 If your image was shot in low-light conditions, you can use 

 to remove the overall RAW noise, then can use 

Remove Object

The Denoise enhancement eliminates distracting noise caused by low-light conditions or file compression while recovering crisp details. It also handles slight chromatic aberration.


It's important to mention that if you shoot RAW, with a camera that has a Bayer sensor, make sure to import your RAW file directly in Photo AI (in standalone or from a plugin that supports sending the RAW file to Photo AI), to benefit another enhancement -> 

This RAW Denoise enhancement is different then Denoise. RAW Denoise will only kick in by importing the RAW file in Photo AI. 

It will remove the RAW noise and hotpixels on this RAW file. Denoise is a different enhancement and is meant for low-light noise, not for RAW noise removal. 

Using Denoise to remove RAW noise may create artifacts and will not allow for hot pixel removal. 

Reach out to support@topazlabs.com with details on your current workflow if you want help on how to tweak your workflow to allow RAW Denoise to kick in; to get the best possible results. 

 tab to select the AI model and adjust the settings. View the 

 article for more info on how to adjust the mask and select different parts of the image.


 for steps on how to activate this specific slider, if needed.

 model to remove low-medium noise caused by low-light conditions or compression artifacts. It is ideal for images from smartphones and cameras with their own denoise models to avoid over-processing.

 model strongly emphasizes retaining image detail for medium-high noise images. Use it for images from very low-light conditions.

 model prioritizes removing as much noise as possible rather than preserving detail. Use it for images with severe noise or heavy compression, such as those shared over the internet.

 slider towards higher values to remove more noise. Moving it lower preserves more detail but allows more noise in the output.

 slider to counteract some of the smoothening done by removing noise. Avoid using this if the Sharpen filter is already on to prevent over-processing.


The Minor Deblur slider was removed from being on by default in Photo AI v4.0.0 and up. To turn it on, go in Edit (Topaz Photo AI menu on Mac) > Preferences > Auto-Pilot > Show Minor Adjustement Advanced Controls > Toggle it on > Save. 

 slider to restore fine details lost during noise reduction. Unlike reducing the 

 slider, the slider brings back some of the original details. Some luminance noise will also return, but not color noise, which is preferred for a natural look.

Denoise Non-RAW

The Sharpen filter produces clear images with crisp details. Unlike traditional sharpening that darkens or brightens pixel edges, it reverses the root causes of blurriness, such as camera shake, motion blur, and missed focus.

 article for more info on how to adjust the mask and select different parts of the image.

 model for all-purpose sharpening. It is intended for images with slight amounts of lens and motion blur.

 model for very blurry and out-of-focus images. Try the 

 model to avoid over sharpening and generating unwanted artifacts.

 model to images affected by imperfections caused when the camera lens fails to focus correctly.

 model to correct the streaked or smeared effect caused by the movement of either the camera or the subject during the exposure time.

model to sharpen objects and keep texture looking natural.

model to emphasize details and bring out finer lines or texture.

 slider to control the degree to which the AI model sharpens the image. It increases the clarity of the details by making the edges and fine features more distinct.

Sharpening can accentuate noise in an image. Use the 

 slider to remove slight noise so the Sharpen filter can enhance the details without amplifying unwanted artifacts.

 slider because it significantly sharpens all details. If there is noise that can be confused as detail, use the 

 (for low-light noise) as a first step to remove it fully before sharpening the image.

The Minor Denoise slider was removed from being on by default in Photo AI v4.0.0 and up. 

, go in Edit (Topaz Photo AI menu on Mac) > Preferences > Auto-Pilot > Show Minor Adjustement Advanced Controls > Toggle it on > Save.

If your image still does not recover details using the Sharpen filter, consider using 

 instead, for missed focus and soft focus cases. 

Sharpen

The Adjust Lighting filter modifies exposure and contrast to produce a well-lit and vibrant image. It makes changes on a pixel-by-pixel basis according to their unique characteristics and surrounding context instead of adjusting the image universally.

This enhancement is a very demanding tool. To optimize the best results, please verify that your computer meets 

When Adjust Lighting is used with RAW images, the output is forced to TIF file format instead of DNG.

Use the dropdown menu to select the area of focus that needs lighting adjustments. Once you select a selection area, the Preview window highlights the area of focus in red. 

Choose between two different model versions in the drop down menu: 

 is a model designed with a focus on delivering a consistent "professional" lighting style across images. It was trained to standardize lighting, often emphasizing a specific style, but lacked the ability to adapt to the unique context or lighting conditions of the image itself. As a result, the output can sometimes feel rigid, as it doesn’t account for variations in lighting needs across different scenes or environments. The model's primary goal is to produce a uniform, professional aesthetic, without considering the diversity of lighting types that may be more suitable for various contexts.

 slider to determine the extent of processing the AI model performs. It does not mean to increase or decrease exposure.

Unlike other editors, increasing the slider value does not necessarily result in a brighter image. It applies more or less of the effect of the AI model.

The default adjustment is 25. A slider value of 0 indicates no adjustment.

Set the slider value between 1 and 25 to have less of an adjustment but in the same direction.

Set the slider to a negative value to adjust in the opposite direction that autopilot recommended.

To change the default adjustment, go to Photo AI Preferences > Adjust Lighting.

offers a more sophisticated approach, producing a more natural and contextually appropriate lighting output. This model is trained to handle a broader range of lighting styles and adjusts its output based on the specific requirements of the image. It not only retains the ability to match professional lighting but also factors in the diverse lighting conditions present in the scene. By considering multiple lighting types, V2 is able to create a more dynamic and visually appealing result, offering greater versatility and a more refined understanding of how lighting should interact with different images.

 -This toggle is turned off by default, so Adjust Lighting will improve lighting while maintaining the existing colors of your image. Turning the switch on will further optimize colors alongside the lighting adjustments.

 - adjust the overall brightness of the image. Increase to brighten, decrease to darken.

 - adjust the brightness of the brightest area. Lowering the highlights can recover lost details, while increasing them enhances brightness and contrast

 - control the brightness of the darker areas in your image. Raising the shadows reveals details in dark regions, while lowering them deepens contrast for a more dramatic look

Adjust lighting v2 is the default model for the Adjust lighting enhancement filter. 

Adjust Lighting

The Balance Color filter neutralizes unwanted color casts and allows creative tone adjustments through cooler and warmer enhancements. It adjusts colors pixel-by-pixel, adapting to each pixel's surrounding context instead of applying universal changes to the entire image.

When Balance Color is used with RAW images, the output is forced to TIF file format instead of DNG.

Use the Temperature slider to shift the image color from cooler (left) to warmer (right). It starts at a default of 50 as a neutral baseline for true white.

Use the opacity slider to dictate how much the AI-generated image replaces the original. The input image is unchanged at 0 opacity. The output is entirely AI-generated at 100 opacity. The default is 100, but if the results appear over-saturated, decrease the opacity to soften it.

Balance Color

Clarity in facial features enhances an image’s communicative value. The Face Recovery enhancement recovers small or low-resolution faces by generating a high-quality version and blending it with the current face.

On file import, Photo AI will analyze the file and look for faces. Then, you can either manually add the "Recover Faces" enhancement or if you have Autopilot on, it will kick in automatically when faces are detected. 

Once the Recover Faces enhancement is added, make sure that all the faces you want to enhance have a yellow selection around their faces. 

Important: The yellow selection rectangle is not showing what area is affected by the enhancement. The yellow rectangle is only to determine 

which faces will be affected by the enhancement

 (yellow rectangle), and which will not (grey rectangle).

 In no way is this representative of the affected area. 

 should be toggled on or off. You can toggle this on/off in your Selection Menu of Recover Faces (available in v4.0.0 and up) or in your Edit (Topaz Photo AI menu on Mac) > Preferences > Recover Faces > Parts to enhance. 

To get the best blend possible, make sure to not only play with the Hair and Neck options, but make sure to recover the rest of the file, 

s. Reach out to support@topazlabs.com with your original and current results attached to your email and we can return steps on how to get the best possible results. 

If a face is not "detected" and not available to be selected in the Selection options of Recover faces, it is not possible to "force" Photo AI to "detect" this face. This can occur if the face is too small, does not have enough details to be recovered, if the face has hidden facial features, is sideways, etc. 

Use the Face Recovery filter on low resolution faces, even if it is large and takes up most of the image. Resolution is measured in pixels.

Use the Face Recovery filter when the faces are small and far away from the camera. The model generates the best results when the face still has sufficient detail. See the full image and zoomed-in portion below.

Using the Face Recovery filter can over-process faces that are already high resolution. This causes a loss of detail and creates a “plastic” feeling because it is too smooth.

Do not use the filter for out-of-focus images. The lack of detail hinders accurate facial reconstruction and generates unrealistic artifacts.

Recover Faces

The Preserve Text filter makes text as readable as possible. It optimizes text presentation by correcting issues such as poor focus, motion blur, small text, low-quality text, and artifacts.

The mask is the first thing that appears when the Preserve Text filter is enabled. It cannot be skipped because it is applied directly to the input image and overwrites the other layered filters.

No text is selected by default. Adjust the brush size and mask the text in the image. Click the Subtract button to unbrush portions of the mask and the Select None button to clear it entirely.

Choose “Select All” for images predominantly consisting of text. Keep in mind the masked area is not affected by other filters.

Use the Low Resolution model to preserve small text. It is best used when the text is a small part of a larger image.

Use the Noisy/Compressed model to preserve text in noisy or compressed images. It removes the noise to preserve the text under it.

: Low resolution reduces image clarity by decreasing pixel count, while image compression minimizes file size by sacrificing some details. The two models are not interchangeable because they address different root causes of poor quality.

Use the Strength slider to control the degree of generation that the AI model will use to preserve the text. The auto-select value will start low because setting the strength too high causes the text to look unrealistic.

Preserve Text

The Upscale filter enhances the resolution of images by up to 600%. It also preserves more intricate details from the original image than other upscaling tools.

 panel to select the AI model and adjust the scale and dimensions. View the 

The AI models rely heavily on existing image quality. They may generate unwanted artifacts if the resolution of the starting image is too low. Upscale from an original image whenever possible rather than a copy that has been compressed.

Make sure to use the appropriate model for your image. The models are not interchangeable and using the wrong model may deliver unwanted results.

The Standard model is the go-to choice for general use unless the image clearly belongs to one of the other categories. It is tailored for medium-quality images from point-and-shoot cameras and older smartphones.

Use the Standard model for highly detailed AI-generated images from Midjourney, Dall-E, and similar tools.

Use the Standard V2 model to have a more generative output than the original standard model. 

Use this model for images that are already high-quality to preserve intricate details. Images from DSLRs and most modern smartphones will fall under this category. Use it for highly detailed AI-generated art as well.

Use the High Fidelity V2 model to intentionally preserve original grain or noise for aesthetic purposes. 

The Low Resolution model prioritizes preserving overall quality rather than fine details. Use it to upscale low-quality images that have been highly compressed.

The Graphics model upscales simple illustrations such as cartoons.

Low light conditions, sensor limitations, or compression artifacts can cause slight noise. The Minor Denoise setting removes this noise before the image is upscaled.

Use the Minor Deblur setting to avoid emphasizing existing blurriness caused by camera shake, motion blur, or a lack of focus when the image is captured.

: Avoid using the Minor Denoise and Minor Deblur if the Remove Noise and Sharpen filters are already enabled. Using both may result in overprocessing.

Visual distortions and artifacts may appear when images are compressed to make them more space-efficient.

Choose from pixels, inches, or centimeters to specify dimensions. Set values for each side or enter the overall DPI (dots per inch) or PPI (pixels per inch). Use this tool to define dimensions for printing purposes.

The pixels/inch and pixels/centimeter textboxes only appear when the dimensions are set to those units.

Upscale & Resize

Functions

Cloud Rendering

Dive into the details of various features of Topaz Photo AI. Learn different elements of the app and best practices for how to use them. 

Enhancements