Can AI SAVE IMAGES?
It seems that all we are seeing nowadays is AI this and AI that, and while the jury still seems to be out on whether this new advancement in machine learning is a good or bad thing we should look at it on a case by case basis to weigh up the pros and cons that could be a possibly for us both now and in the future
There have been a number of advances recently, which in particular to photography have the potential to be game changers! The first of these seems to be the ability for AI to look at a specific image and apply its magic to make changes that may not be easily possible for a meat human. Let’s look at one in particular, AI noise reduction, which promises the ability to reduce/remove noise from an image without comprising on image quality. To do this we should look at how noise works or doesn’t work in our images.
Noise refers to the random variations in brightness and colour that can appear in an image. It often appears as tiny specks or grain-like patterns and can be distracting or undesirable, especially when it significantly affects the quality of the image.
Noise is primarily caused by the camera's image sensor when it tries to capture and process low levels of light in a photograph. When there is not enough light available, the camera amplifies the signal from the sensor, which can introduce random electrical fluctuations. These fluctuations result in the appearance of noise in the final image.
Various factors can contribute to the amount of noise in a photograph, such as the camera's ISO setting, the exposure time, and the sensor's quality. Higher ISO settings, longer exposure times, or using cameras with smaller sensors can increase the likelihood of noise.
Photographers often try to minimize noise by using proper lighting, avoiding high ISO settings when possible, and using noise reduction techniques during post-processing. However, it's worth noting that in some cases, noise can be intentional and used creatively to add a certain aesthetic or mood to a photograph.
Thankfully we can reduce the appearance of noise in our images with Lightroom or Photoshop
In the Reduce Noise dialogue box, you will see several sliders and options:
Strength: Adjusts the overall strength of the noise reduction effect. Increase it to reduce more noise, but be careful not to overdo it and lose too much detail.
Preserve Details: This slider helps retain important details in the image. Increase it to preserve more details, but again, be mindful of not introducing artefacts.
Reduce Color Noise: Controls the reduction of colour noise specifically. Adjust it based on the amount of colour noise present in the image.
Sharpen Details: This option helps restore the image's sharpness after noise reduction. Use it to bring back some lost detail.
While adjusting the sliders, keep an eye on the preview window to see the changes in real time. You can zoom in and pan around the image for a closer look.
Once you're satisfied with the noise reduction settings, click OK to apply the changes.
It's important to note that completely removing noise without affecting image details can be challenging, especially if the noise is severe. Therefore, finding a balance between noise reduction and preserving image quality is crucial.
Additionally, Photoshop offers other noise reduction methods and plugins that you can explore for more advanced noise reduction techniques, one of which is Noise AI
My Original Edit, Which was full of Noise
Noise AI is trained using a vast amount of image data that contains both clean images and their corresponding noisy versions. These noisy images serve as input to the AI model, while the clean images act as the desired output. By comparing the two, the AI algorithm learns to recognise patterns and characteristics of noise in different types of images.
Noise Analysis: When you upload an image to Noise AI, the software first analyses the image to identify the presence and type of noise. It examines the patterns, textures, and variations in brightness and colour that indicate noise artefacts.
Noise Reduction: Once the noise analysis is complete, the AI algorithm applies its knowledge and techniques to reduce the identified noise. It does so by selectively removing or reducing the noise while attempting to preserve important details and sharpness in the image.
A close up of some of the noise Challenges
Adaptive Filtering: Noise AI employs adaptive filtering techniques that adjust the strength of noise reduction based on the specific characteristics of the image. This adaptive approach helps avoid over-smoothing or excessive loss of details.
Post-processing Options: Noise AI typically offers various post-processing options and customisation settings that allow users to fine-tune the noise reduction process. This includes adjusting the level of noise reduction, preserving certain image features, or applying additional sharpening or detail enhancement.
Output: After the noise reduction process is completed, Noise AI generates an output image with reduced noise and improved image quality.
Overall, Noise AI combines the power of artificial intelligence, machine learning, and sophisticated algorithms to effectively understand and reduce noise in images. By leveraging its training on a diverse range of noisy images, the software can intelligently analyse and enhance the quality of images, providing cleaner and more visually appealing results.
After AI
I had an image taken several years ago of a scene that was extremely dynamic during a red warning storm off the coast of Ireland. I was at Clogher Strand and while looking out to sea towards the sleeping giant ( An Island that lies off the coast) the incoming waves were smashing against a rock that sits in the bay, producing some incredible wave explosions with almost every wave that arrived. The challenge however is the scene is typically darker as there are thick dark storm clouds, and you need a pretty fast shutter to freeze the wave and preserve the detail.
My shutter speed was good, however, my ISO wasn’t, and the resulting image was a darker-than-ideal shot which was full of noise. Time to put this new AI tool to the test and see what it could do with my image.
You can watch the editing process in the link below
What are your thoughts on this topic? Let’s continue the conversation in the comments