iStockphoto Photography Standards: Noise and Compression

Noise (or pixel discolouration) and compression artifacting are two different technical issues that photographers face. In both cases, an imperfection is introduced into your image at some stage during your workflow. Depending on the severity of these imperfections they can limit the usefulness of your image for stock.

The good news is that improvements in technology are gradually making these problems less common and serious. However even if you are still using older equipment you can still learn how to minimize the problems.

We will roll this article out in two parts, dealing with Compression first and following later on with more on Noise. This article is primarily geared towards beginners but we will get into some more advanced examples as we go. The images were created especially for the piece. In most cases they have been pushed pretty far in order to illustrate the problems. Compression does not need to be as severe as the examples here in order to cause a rejection from iStock.


Compression artifacting is, after lighting, probably the most common reason for rejection here at iStockphoto.

Compression, and 'Compression artifacts', refers to visual distortion which occurs in an image when information is lost. The JPEG file format literally compresses an image - makes it smaller - to reduce the file size during saves. The particular compression method it uses is a lossy format, which means that it loses some information to shrink the actual number of bytes used in the file. When too much information is lost it can have a visible impact on the image. This can also happen during the image editing stage.

Compression typically results in a few different kinds of visual distortion, which will typically be called 'artifacting', 'contouring', or 'posterizing':

• Curved edges in detailed areas take on a jagged stair-case like appearance.

• Checkerboard style blocks appear.

• Color gradients become 'banded', meaning that instead of a smooth transition between colors, there are instead abrupt changes from one color to another making visible bands across the image. This often happens in large areas of subtle color change, like a clear blue sky in a landscape.

This crop of a blue sky is heavily compressed. The gradation between blue shades is visibly banded and what should be smooth areas of colour are instead severely posterized.

This is an extreme example: an image can suffer from too much compression without having to look this bad. Watch for splotchy, blocky areas in what should be expanses of solid colour, and visible borders instead of smooth transitions between colour tones.

Once an image has these problems they are nearly impossible to fix. Sometimes people will spot compression in an image and try to hide it with additional editing — either through some sort of noise-reduction application or with selective blurring. These tend to introduce other problems though as any 'fix' will still look unnatural and will merely have degraded the image quality in a different way.

Compression happens to images for two main reasons: either due to limitations with your equipment, or as is more and more commonly the case, as a result of overly-aggressive editing and processing. In both cases preventing compression comes down to your workflow.

1. Use a DSLR

There are a lot of different compact digital point & shoot cameras out there. In general they just don't offer the detail quality of a DSLR. Many compact cameras have too many megapixels crammed into too small a sensor, and employ built-in in-camera JPEG processing with no manual override settings. These can be introducing compression problems into images right off the hop. If you are serious about producing stock images, forget the compact point & shoot and get a DSLR. They get cheaper all the time and there are plenty of different models to choose from.

These two examples suffer from classic "Sub-compact" compression. They were shot with point and shoot cameras that just weren't up to the task of accurately reproducing detail.

The whole scene in this image is severely compressed and appears to have been made from building blocks.

2. Shoot RAW

Unlike JPEG, RAW is a lossless format, meaning that information is not lost during the process. By shooting in RAW and putting off saving to JPEG until the final stage of the process, you will conserve as much information as possible throughout your workflow. Traditionally the main drawback of the RAW format has been the larger file size. These days however disk storage, compact flash, and other cards are continuing to come down in price all the time. At the end of the day, it is not worth putting up with severe jpeg compression in order to cram more shots onto one card.

Original RAW file.

Converted JPEG file. Conversion shows several RAW editing decisions in action, mostly regarding contrast and colour correction. Looks good at this size, right?

100% crop of the above example. The editing has produced some visible compression, detail degradation and blocky posterization.

The main advantage of the RAW format is that you never lose the original information and can always go back. If artifacting happens during editing or saving, the original file is still there and you can start over without completely losing the image. So in the above example, even though we've introduced compression into the image with our editing, we can still go back to the undamaged original and try again.

Compression vs. Overfiltering

As we mentioned above, most of the compression we see these days is caused by excessive image editing. Pushing certain filters or processing techniques will cause information to become lost and introduce compression damage into the file. We've prepared some examples of this kind of editing damage:

Original RAW image, taken with neutral density and colour correction gradation filters.

Converted JPEG file, 100% crop. Clear compression damage at this size: jagged lines on the outlines, over sharpened, saturation overkill.

100% crop showing blocky compression throughout the sky.

Edited RAW file where we've attempted to correct the shadows and highlights by bringing them up significantly.

This 100% crop from an area of the sky just underneath the cross shows just how badly compressed the detail has become.

Edited original RAW thumb of abandoned underground WWI bunker. Shot at high ISO because of low light conditions, resulting in a very toned look.

Converted jpg file, 100% crop; out of focus wall, front right side.

Converted jpg file, 100% crop; out of focus ceiling part at the back of the corridor.

In both example crops the issue is that processing the high ISO original has pushed over the border between acceptable "grain" into visible compression, where detail is smudgy and much of the definition is lost.

Original RAW file, unedited

RAW file edited to bring back detail in lost darker areas, curves adjusted for warmer tones, contrast and brightness selectively edited, colour (chrominance) de-noised, etc.

JPEG conversion, 100% crop

JPEG conversion, 100% crop

100% crop of original raw file, underexposed, dull colours

100% crop of edited raw file, selective exposure correction & curves adjustment resulted in visible detail degradation

Preventing Post-Processing Compression

1. Be gentle

If you are using JPEG, always proceed carefully. Every time you make a change to the file there is a chance that you will introduce compression problems and its important that you can go back if you make adjustments that damage the file.

Basically, when working with an 8-Bit JPEG file any severe editing of the contrast, saturation, colours, or detail sharpening can introduce artifacting.

Suppose you have a beautiful blue sky in a landscape shot for example. You may be tempted during editing to push the saturation and contrast a bit over the top in order to get a vivid and bright colourful image with a wow effect. Check the sky again though – there is a chance you've introduced visible and unpleasant posterized banding on the otherwise smooth gradient between various blue colours.

Extensive curves editing can bring similar results. Even RAW -> JPEG conversion, when abused, can compromise your detail.

• Always save your JPEGs with the maximum quality settings.

• Convert wisely and gently.

• Edit gently as well in a non-destructive fashion, using new layers when possible.

2. Convert for substantial editing

If you do need to push something hard during the editing stage, try converting to 16 bit TIFF and edit in that mode. Editing in 16 bit mode lets you use a lot of extra information before you dump the image down to the 8 bit JPEG.

3. In-camera preparation

Because those blue skies can be problematic, do what you can during the shoot to capture the necessary detail and avoid compression altogether. A polarizing filter can be your best friend for those beautiful blue summer skies. There are other usable filters that will improve your original landscape shots – colour correction filters, neutral density filters, or a combination of both as gradation filters.

Look closely before the final save

Calibrate your monitor, use profiles, and stay within your working colour atlas. This will all help you see your image during the last 100% corner to corner inspection before your final save. Compare the final with original to make sure no detail issues have been introduced.

Coming soon: Noise

In the meantime, check out our previous Photography Standards articles:



Model Releases

3D Renders

Prague Cityscape image by emigee.
All other example images were created by Gremlin.
Header image: Television Static by kyoshino

Matjaž Slanič (gremlin) is iStockphoto's Technical Lead.

Andrew Wedderburn (rogermexico) is iStockphoto's Community Manager.

Not a member?Join