Raw File Data Collection
Here on f.64 Academy and f.64 Elite, I talk about data manipulation quite a bit. More specifically, I show you how to edit your images in Photoshop (data manipulation). I have done well over 300 tutorials on data manipulation in Photoshop, but I have been remiss about a critical element to the process. Not all data is created equal!
When I am on location, I consider myself a data collector. I am using my camera (it’s sensor) to collect the data from the scene to use it later in my post-production efforts. I understand this process, and the importance of the data that I collect, because I have done countless experiments with the received data.
So where have I been remiss?
I haven’t shown you all of these experiments and assumed that you know about this sort of thing. In today’s tutorial, I want to take the time to show you how I analyze data, make judgment calls on tradeoffs, and empower you to experiment with your own data collection device. This is probably the most crucial step in the whole process of image editing, so please take the time to experiment with your sensor as you may see different results than the ones in this tutorial.
What I have noticed with my Sony Cameras (a7rII and a7rIII):
- When shooting, I tend to shoot more to the left of the histogram as I noticed it handles noise pretty well. I can get away with a lot of pixel pushing in the dark areas.
- I noticed that when I increase the dark areas, I do get noise (not unbearable), but it comes with a green color shift in the shadow areas. Easily fixable on both accounts.
- I noticed that the Sony’s do not handle highlights as well as other sensors I have used. When pushing the RAW data with a heavy positive exposure, I find highlight recovery to be less than desirable. A lot of blown areas.
- Also, there seems to be a Cyan color shift in blue skies when heavy positive exposures are reduced in post.
What do you notice about your camera sensor when manipulating the Raw data?
Blake, great thought process that I have not considered. I will check this out on my Pentax K3’s and K1 Mark ii sensors. Way to go as usual Blake! Thanks,
Terry B., GP, Oregon
Not to make this a Pentax fest, but I have found much the same with my K-3 and K-1 as well. Pentax mount Sony sensors of course and are reputed to have developed an image path such that the IQ of a Pentax image is incrementally superior to a Sony body with the same sensor. I usually ‘expose to the right’ to preserve highlight detail, safe in the knowledge I can push my raw files at least 2-3 stops without much fear of degradation
Errr, I meant ‘expose to the left’. Where is the emoticon for chagrin?
Perhaps this explains why I have developed confidence in low-light shooting situations. I shoot Sony A6500 and the Sony RX100, which is my “purse camera.” I have much better success bringing to-the-left images to a brighter end than the opposite. I never noticed the cyan-shift until you demonstrated it in the video. I’ll keep that in mind.
Hi Blake! I have followed the teachings of bird photographer Arthur Morris for a long time. To get the “best” exposure “every time” he recommends ETTR (expose to the right). Expose until you get a few “blinkies” (highlight alert) then come back one click. This advice was for Canon cameras. I have an old 7D and this works like a charm to get rid of noise. By today’s standards the old 7D is VERY noisy. However in recent times I have noticed that my skies are VERY cyan and I have been re-processing a lot of my older images. Now I have an explanation for why I have been re-processing images in recent weeks. Noise in my darks will still be the main killer in my images until I upgrade my gear. Many thanks for this info.
Hi Blake – call me nuts but I’ve been experimenting with placing my af point over a highlight I want to have detail, activating the AE Lock, placing the af point over my focus point, locking focus, composing and taking the shot. It places the metered highlight at midtone – I compensate by adding some exposure to taste. I’m finding my images sharper (allows using a higher shutter speed) – colours are better (not pushing exposure to the right) and I can pull detail and contol noise in the shadows (if needed to brighten these areas) – generally shadows are not the subject its discretionary. I’ve been experimenting with this in camera (I’m using a Fuji X-H1).
Very interesting. Somehow I knew I was reading a comment from a Fuji shooter. I do exactly the same with my XT-2.
Most illuminating! I use a Nikon D800 and a Nikon D7000. I have never really thought about shooting to the left or right of the histogram except noticing better data collection to the left on my D800. More experimentation needed. Thanks, Blake
With Canon, you have to be careful with the shadow values. I test out the dynamic range of my bodies. Then work out the difference between a meter and my bodies meter. Use FastRawViewer to see RAW Histogram, which is accurate untouched info. In ACR using V2 engine, zero everything and just use the exposure slider to brighten or darken the file before opening in PS to do all tone, colour, contrast and effects.
FastRawViiewer and RawDigger are very useful if you like RAW data info. Use them to take into account the following,
love to hear your views on digging a bit deeper
Hi Blake. Nice one! I realise this is what I do at the taking stage without really thinking about it through. The explanation will now make me take my images in a more considered way. When I slip out to take some photographs in future it won’t be called that. I will now be photographing nature’s data. It makes me sound much more serious about my hobby!
You are doing it wrong. Lightroom & ACR automatically adjust RAW images by moving the centre point of a curve, according to what their algorithm thinks is 18% grey. Try it with their 2010 process and it will be corrected – but you will lose some good things as well!
You should try Sony’s own ‘Imaging Edge’ software – much better.
Very, very interesting. Thank you for sharing.
I suspect the performance variance is connected to a particular sensor’s ISO Invariance factor. The Sony A7 series, and from what little experience I have with the A9, is probably the most invariant of any sensor so that pushing 1600 to 6400 will yield basically the same result as shooting at 6400, which means for color fidelity, getting a better exposure to begin with doesn’t come with as much of a penalty when pushing shadows as it might for other kinds of sensors. Interesting tests. I’ve seen similar color shifting in my drone on jpegs but of course, shooting RAW you get down to the base performance capabilities of a sensor. With my Olympus micro-four-thirds sensors I tend to adjust the overall exposure dependent on the subject matter, so for some, I’ll under expose because I’m really wanting to keep the shadows and get the best color in the highlights I can. For subjects where midtowns are going to be the most important I will push exposure to the right a little, not too much.
I get the RAW data right an left thing and I have found that I will lean to the left also using my 5Dll. One thing I must ask about this is this,”What about the lens?”. Sometimes for fun I stick an old Nikkor 20mm on my canon and the coatings of the lens produce a totally different color capture. Also what about “in camera” settings such as contrast, white balance, etc?
Very thought provoking as usual.
Another great video! As you may have, I found the changes in color depending on the exposure very interesting. With my Canon 5D Mark IV I enable the “Highlight Alert” feature which shows the blown our areas as red on the LCD screen. The Highlight Alert is based on the JPEG generated for the LCD; therefore, I have more data to manipulate in the raw file. When I’m required to hand hold and don’t have the luxury of brackets for HDR purposes, I then a blown-out scene by 1.5 to 2 stops to capture the highlights. The underexposed areas are easier to bring back in LR and PS than the blown out areas. This is a practice I also used when shooting slide film when I didn’t have the ability to check the image on site.
Really interesting stuff Blake. Your thoughts on ‘data collection’ and the conclusion that choosing to under-expose is usually preferable to risking blowing out part of the image is similar in so many ways to how I was brought up to use slide film.
We were always taught that, if in doubt, it was preferable to under-expose at maybe a third or a half stop. Many landscape photographers shooting transparency stock would set their cameras up to permanently under-expose every frame. There was a dual benefit in shooting slightly below ‘optimum’ exposure – the corollary with digital cameras is quite rightly that highlights are less likely to blow out, but slightly under-exposing slide film also leads to increased (colour) saturation. Couple these techniques with a graduated ND or a polariser and you were doing all the things in camera that we now take for granted to be able to accomplish in post in Photoshop/Lightroom.
Even though my knowledge of digital post-production increases with every book I read or tutorial I watch, when I’m out actually taking the photographs (data collecting!), I find I’m still sub-consciously applying the low margin for error practices that were so valuable when shooting slide film where you really had a tolerance of just a fraction of a stop either side to get the exposure right.
As you rightly state, all RAW files are different and vary from camera to camera. Similarly, all transparency stock reacted differently to how much under-exposure you could get away with. It was a trial and error approach that paid off when you could be confident that you’d found a particular slide film’s ‘sweet spot’. For example, Kodachome 25 and Fujichrome 100 always seemed to produce excellent saturation at around ‘correct’ exposure – hence under-exposing just made the scene look too dark. Kodachrome 64 and Agfachrome 100 however seemed to become beautifully saturated at between a third and two thirds of a stop under. My favourite for under-exposing was the criminally under-rated Kodak Ektachrome 100 – you could easily get away with a full stop under, and if the scene was fairly evenly lit, maybe a stop and a half.
I sometimes wonder if we have it too easy these days with the latitude RAW manipulation allows us, but having said that, I still believe there’s no substitute for getting as much right ‘in camera’ as possible.
(also posted on your YouTube channel)
Love your knowledge, and tutorials!
The word “data” is plural!
I have often bracket simply to ensure I get “something” out of a shot. But I have to admit I have not as yet “studied” my cameras to the point to understand their strengths and weaknesses sensor-wise. I will now on my 5DS, 5D4, M5s and M50! Thanks for this!