Lawrence Ripsher

RAW and exposure latitude in post processing

[Note – the samples below may vary significantly depending on monitor type / brightness – the final example looks v well exposed on my monitor. Your mileage may vary on your own setup)

This time round I’m going to touch briefly on exposure, RAW and post processing.

One of the major benefits of a good sensor which produces image quality at high ISO is not so much the lack of noise that the sensor produces, but that the image retains good detail, dynamic range, contrast and colour. Too often noise is measured as a lack of noise – and camera companies occasionally latch on to that and try to get round the issue by applying heavy post processing techniques. The result? Images which are relatively noise free but end up having a pale, washed out, water colour like effect which is instantly recognisable even at web resolutions.

The thing is, the presence of noise does not detract from an image (look at good film grain @ iso 1600 and you’ll know what I mean) – it’s more the fact that it ruins attributes associated with image quality.

Related to this is the fact that a good sensor gives you latitude for errors and post processing, even at high ISO’s. For advanced and professional photographers, the ability to be able to successfully manipulate / mould an image into a finished product is sometimes as or even more important than what initially rolls off the compact flash card in the first place. In this post, I’m going to show how the 5D Mark II performed in such a situation – the scenario is an example of when the photographer (in this case, me) screwed up a shot at high ISO and whether the camera provided enough latitude to to rescue the image. Not your usual High ISO / Noise tests I know – but in many respects just as relevant (and besides, the regular “here’s a shot @ iso 3200” will come later).

So here’s the shot, straight out of the camera:

The shot settings;

- Lens: 35mm f/1.4

- Settings: 1/4000s, ISO 1600, f/4

I know… what the hell happened? Well I was trying to catch my dog in motion as he was destroying a snowball mid air. The camera was set to Aperture mode and matrix metering and metered too much of the snow and underexposed the image. This is typical when you shoot a mostly white background (sky, wall, snow, etc) and I hadn’t compensated for it or switched to manual as I normally would (it was early in the morning). Still, I knew my timing was good and wanted to see if I could rescue the image in post.

The problem? Well the image was at ISO 1600. This means that noise will be present in the image but whenever you try to correct underexposure in post, it significantly increases the amount of noise in the image. So, an even better test.

So here’s what I did:

- Open the image in Canon’s DPP RAW converter and apply a significant curve boost to push the midtones and shadows

- Convert the image to TIFF (8 bit)

- Open the image in CS4. Apply some additional levels tweaking, increase shadows now to correct the overexposure and take a burn tool to apply it to the snow (which the curves boost had caused to almost disappear). Zoom and crop for better framing

- All this work started to do funny things to the colour of the snow which was turning blue. So I quickly took a desaturation brush and removed the unnecessary cold colour of the white balance

- Dodge his face just a little, then resize, sharpen and post to web

(it sounds like a lot but this took about 2 mins).

And finally, the finished result:

In rescuing the image I actually realised could have focused a little better on his face – but hey, the timing was good so not bad for a first attempt at before 8 in the morning.

So how did the camera do? Well the RAW image that rolled off the sensor, even at ISO 1600 provided a pretty incredible amount of latitude for post processing. Being able to do this at a low ISO on the previous 5D wouldn’t have surprised me – but being able to do it at ISO 1600 was fantastic. Even with the massive underexposure, I was able to boost the shadow details significantly – all the while being able to retain an acceptable amount of the colour, contrast and dynamic range of the scene (remember, black dog on white snow – doesn’t get more demanding than that from a DR point of view).

With the original 5D, anything this underexposed I would typically have thrown away at ISO 400 and above. Some might remark that the best solution is to just expose correctly in the first place – to which I’ll wholeheartedly agree. But it’s like insurance, you never really care about it until you need it and besides this extra cushion opens up an exciting new number of opportunities for pulling details out of shadows – and obviously not just at high ISO. In essence, you can consider that to be the real point of this post.

So there you have it – an unscientific but very real world example of what good high ISO performance can give you. More topics coming soon…

Related posts:

1. EOS Canon 5D Mark II and Olympus OM Zuiko adapter I've been building a collection of Olympus OM Zuiko lenses...
2. Zeiss 85mm f/1.4 ZE Planar T* Lens (for Canon EOS) Review The name Zeiss needs no introduction in the field of...
3. Olympus OM Zuiko 135mm f/2.8 Lens Review (with Canon EOS) I recently wrote about an adapter I bought for...
4. The Dynamic Range of Film A follow up to The Resolution of Film. These are...
5. Olympus E-P1 Review Diary – Part 9 – Art Filters and Samples Images V I'm not generally a fan of art filters / in...

Related posts brought to you by Yet Another Related Posts Plugin.