Balancing Exposure and Processing

[J-see]

As much as I applaud your courage to try proving this theory, I'm not convinced of the value of this. Of course, you can recover a raw file, but getting the sensor to "grab" more light in the shadows specially is a huge plus for me. Now time might prove me wrong, but where I come from (film days), when you have nothing on the negative, you're left with nothing on the print.

I'm not disagreeing here; if shadow detail is important, ISO might do a better job but in reality, the gain in the shadows is very limited and flatlines at a certain ISO. The actual difference between ISO 400 and ISO 12500 in shadow improvement for my cam is that small, it's almost ridiculous.

Btw, what I do is almost exactly like negatives; those too had to be exposed before there was something to print.

 

[Marcel]

And something else to remember, all sensors were not built equal. The D750 sensor might be a lot better than the D700 sensor for example, so the results you are getting might not apply to all other cameras... Food for thoughts.

 

[J-see]

And something else to remember, all sensors were not built equal. The D750 sensor might be a lot better than the D700 sensor for example, so the results you are getting might not apply to all other cameras... Food for thoughts.

Certainly, sensors differ and thus anyone desiring to find out has to check the information and test the limitations.

My cam doesn't have shadow improvement up to around ISO 300 and after 400 gains that little it simply doesn't justify the loss at other levels.
The D700 has zero improvement up to 360 or so, then gains quite a bit and half that more at 800 and then flatlines until the end.

But even this information isn't enough to base one's shooting upon. Sensor pixels, conversion noise and the full well capacity have an effect and how the actual ISO increase is applied in cam also makes a difference.

This is not a one-for-all method of shooting; it has to be individually tested to see when it benefits and when there's no advantage.

It's a different style of shooting but not one that will suit all. It's one that costs quite some effort in the beginning but after that, pays the investment back.

 

[J-see]

Here's an example without the color shift. I fine-tuned them in PS.





More extreme:





And I'm done testing and adjusting. Now I can apply it and focus on the shots.

If anyone wants to try this technique and needs some information, gimme a yell.

 

[J-see]

More on processing:

I load the shot as is into LR and adjust the black clipping point, cam and lens profile. Then I go to PS and use levels first, then curves, to use the information available to spread it correctly. I don't even fine-tune, I just select one of the options.





Not even half a minute of work in PS.

Here's one processed in LR after an awful lot of work:



Then the same shot reprocessed from scratch in PS first applying levels/curves, then exposure and highlight/shadows. It took me a fraction of the time of the other in LR.



Tomorrow I'm to the store ordering PS for Mac.

 

[Eyelight]

I reckon I don't grasp what takes so long to do in LR. I don't always do the same exact flow, but unless something is way off and I'm trying to recover, I probably spend about a minute or maybe two.

I suppose though that processing is definitely a "to each his own" thing, which is the same as a lot of photography.

 

[J-see]

I reckon I don't grasp what takes so long to do in LR. I don't always do the same exact flow, but unless something is way off and I'm trying to recover, I probably spend about a minute or maybe two.

I suppose though that processing is definitely a "to each his own" thing, which is the same as a lot of photography.

The problem was getting the colors right. LR is very crude at that level and for the RAW+ files I shoot, it takes some time playing with curves and adjusting contrast to have them reasonable correct. It was the last issue I was struggling with.

While trying to fine-tune some I decided to go PS and discovered I can get it right there within seconds.

Only if I am geeky about certain aspects I'll waste loads of time in post but the other shots will be processed much faster now I know how to normalize them in PS.

I've been shooting the D3300 using the same approach. I'm curious how that'll do.

 

[Eyelight]

The problem was getting the colors right. LR is very crude at that level and for the RAW+ files I shoot, it takes some time playing with curves and adjusting contrast to have them reasonable correct. It was the last issue I was struggling with.

While trying to fine-tune some I decided to go PS and discovered I can get it right there within seconds.

Only if I am geeky about certain aspects I'll waste loads of time in post but the other shots will be processed much faster now I know how to normalize them in PS.

I've been shooting the D3300 using the same approach. I'm curious how that'll do.

OK. I didn't realize it was connected to the low ISO approach.

 

[hark]

Here's one processed in LR after an awful lot of work:

View attachment 136446

Then the same shot reprocessed from scratch in PS first applying levels/curves, then exposure and highlight/shadows. It took me a fraction of the time of the other in LR.

View attachment 136447

Tomorrow I'm to the store ordering PS for Mac.

I prefer the first one. In the second, the spots on the top duck compete with the background more than they do in the first one. Would have liked to see the exposure on both ducks raised a little without lightening the background. Just my opinion...my taste may not be the same as yours.

 

[J-see]

I prefer the first one. In the second, the spots on the top duck compete with the background more than they do in the first one. Would have liked to see the exposure on both ducks raised a little without lightening the background. Just my opinion...my taste may not be the same as yours.

I didn't pay that much attention to it but the PS version indeed makes the second duck blend too much into the background. I should have adjusted mid-tones or their contrast some more or minimized the shadow adjustments of the background. I was paying most attention to the male duck which did turn out better in version two.

 

[J-see]

Here's one taken with the D3300 today. I use auto-ISO up to 400 because it directly gains after 100 and reaches 2/3th stop until there. After 400 the costs do not justify the increase in ISO.

The D3300 evidently doesn't handle noise that well and light was even worse than the shots I took with the D750. But it shows with this cam too it is possible to shoot like this. It only needs some fine-tuning.



That's the darkest I shot today with the D3300. I didn't go as extreme as with the D750 and use up to two stops ISO but even within those limitations, it does well. It never was a good low-light performer.





Noise reduction used in LR.

 

[J-see]

It has been raining the rest of the day so I decided to invest some more time in comparing rather similar shots. Here are some of the results, I normalized the exposure and adjusted everything identical for them in PS with clipping set to 0%. These are shots through quite some ISO range. I'll post the 100% crops in the next reply.

 

[J-see]

100% crops (highlights and shadows killed to open up:















I think we can agree I'm at the very least not losing any quality.

I'd say with the D750 you can shoot almost anything at ISO 100 and do the rest in post.

 

[J-see]

This is the most extreme I managed. It's not that I was trying but it just so happened. I had my ISO at 160 because I knew the light was too low and every shot would be in the high ISO range. I have about 15% of my shadow gain at 160 at the lowest cost possible. 15% might seem low but if you consider my max gain is only 2/3 stop, one could even wonder why I bother.

SOOC:




Seriously processed crop after adding 6 2/3 stop exposure which combined with the initial ISO would range the shot somewhere between 12800 and 25k.



100%, evidently noisy.

 

[J-see]

I had to kill time so I decided to push the D810 and check how it performs when using ISO vs post adjustments. Since the cam uses a double amplifier, it is a bit different than the D750 and I need to test it all. The results are not surprising when it comes to noise but at lower ISO it did make me frown.

I shot all four once at correct exposure using ISO and the other at the same A/S but at native ISO 64 and adjusted exposure in post. No processing done besides cropping, WB and exposure. All 100% crop of shots in reasonable low light conditions.

ISO 800 vs 64 post-exposed:




ISO 1600 vs 64 post-exposed:




ISO 3200 vs 64 post-exposed:




ISO 6400 vs 64 post-exposed:




The 800 makes me wonder what's going on.

Apparently the chained amplifiers work differently from 64-1000 and >1000. That gives me 4 stops I can shoot at native ISO and if I need more, I best use the cam's ISO since it'll do a cleaner job than I can do in post.

If I compare with some shots I took with the D750 at higher ISO, it shows that it is cleaner than the D810.

 

[J-see]

I wanted to check ISO vs non-ISO a bit better for the D810 and shot two versions. Too bad I had no tripod with me but that's for next time when I check how far this works.

I took two shots with the same shutter and aperture, one at native ISO 64 and the other the required ISO for "correct" exposure. I then loaded both into Capture NX, increased the 64 by 3 2/3th stop and saved both as a TiFF. I also loaded both in RT and adjusted the 64 ISO by 3 2/3th stop. Then I adjusted the WB of all by selecting the exact same spot, saved and cropped some parts to compare. There's no other processing done. No noise correction, no sharpening.

Capture processing


vs


Details:







RT processing:


vs


Details:

 

[J-see]

At the level of noise it makes close to no difference. The RT rendering of the 64 is a bit strange and I'll have to check why it is like that.

But this isn't about noise. The main difference is data but it's hard to see that in an 8 bit JPeG.

Even when the shots here look identical, the difference is in the RAW data.

ISO 64
Dynamic Range: <14,76Ev
Tonal Range: <9,83 bits
Color Sensitivity: <25,7 bits


ISO 800
DR: <11,67Ev
TR: <8,28 bits
CS: <21,8 bits

The same shot, about the same effort and without ISO I can have 1.5 times more distinguishable gray levels and more than 8 times as many distinguishable color values. And my dynamic range increase.

All I have to do is NOT use ISO.