For a very long time I have been an advocate of shooting RAW files and processing them through Adobe Camera RAW. Recently I have been shooting a lot of pictures in dark spaces with poor light and have been regularly using ISOs of 6400 and above. What has shocked me is that my normal workflows have been producing results that I find a bit grainy – even with the superb high-ISO capabilities of the latest Canon R series cameras.

Having edited some of the live jpegs transmitted during events it has become apparent to me that I need to get stuck into some of the latest noise reduction options as well as having a better look at what in-camera processing can do.

Following on from a recent job I used several different and relatively unrefined** techniques to produce files that I might send to the client after some minimal colour correction. The worst option, when viewed at 100% was the file produced from the RAW file with standard sharpening. The remaining options all have advantages and disadvantages and which technique is best depends on how long you have to get the work done, what the client wants to do with the files and your own personal taste.

• Jpeg straight from the camera
• Jpeg produced using enhanced AI noise reduction in Adobe Camera RAW
• Jpeg straight from the camera with additional sharpening
• Jpeg produced from the RAW file with standard sharpening
• Jpeg produced from the RAW file with additional Topaz AI noise reduction
• Jpeg produced from the RAW file with no sharpening
• Jpeg produced from the RAW file with manual noise reduction and sharpening

Based on my own experimentation and without getting into the finer nuances of the different workflows I have come to the following basic conclusions:

Straight from the camera – this is, by any standard, a fast and simple workflow. The images are close to being noise-free at 6400 ISO if you get the exposure correct and don’t have to mess about with shadows, highlights or white balance.

Enhanced AI noise reduction in Adobe Camera RAW – really nice results but the workflow isn’t particularly easy given that you have the extra stage of having to produce and then convert from a DNG file. I wouldn’t want to have to do this on a quick edit of 50 images but for one or two frames it is a great way to work.

Straight from the camera with additional sharpening – quick, easy and produces surprisingly good results as long as you don’t need too much other correction.

RAW file with standard sharpening – this is probably the worst option and I am going to have to find a better way to work than this unless the images need a lot of shadows, highlight, exposure and/or colour correction.

RAW file with additional Topaz AI noise reduction – really excellent results. Different to Denoise AI in ACR but equally as good and maybe better for many people’s taste.

RAW file with no sharpening – pretty much identical to the Jpeg straight from the camera but allows for slightly more toning and correction.

RAW file with manual noise reduction and sharpening – for speed and quality this is a decent compromise and represents the most efficient way to do batches but it can’t match the AI options.

Confused? I am. Sharpening has always been the problem when working on high-ISO images. We’ve all known that for a quite a while now but the quality of in-camera processing and the reduced need for sharpening with the current crop of mirrorless cameras has increased the number of workflow options available. As I mentioned near the beginning of this piece, the relative complexity of the workflows with both Denoise AI in Adobe Camera RAW and Topaz DeNoise AI make them the front-runners when you have time and/or a small number of frames. For large batches then reducing the degree of sharpening from RAW files is still a decent option but the end use of the pictures will always dictate which path you go down. It probably comes as no surprise but the basic conclusion is that the longer it takes to achieve a result, the better it gets.

This isn’t the end of my investigation into this complex topic. It isn’t even the beginning of the end but, to continue to paraphrase a famous Churchill speech, it is the end of the beginning. I’m going to work on speeding up my use of and expertise with the various denoise options as well as having a closer look at what others are doing and speaking to clients about their needs.

That’s the thing about this job; you can get stuck in your ways and you can end up getting left behind if you don’t become familiar with the latest tech. It wasn’t that long ago that I would have been having to shoot with flash or putting up with slow shutter speeds on these jobs. Technology has rattled along at a fast pace.

My thinking so far is that this isn’t generative AI so there aren’t too many ethical issues but I’m happy to engage in conversation with anyone who thinks that I’m wrong.

** When I say relatively unrefined I mean that I didn’t spend hours playing about with various sliders and controls but employed basic ‘out-of-the-box’ settings to make this baseline assessment.