dtaylor said:
jrista said:
I usually print 13x19" at home, and I usually do spend quite a lot of time working the shadows to get them to print nicely.
Trimming most of your post because...how on Earth do you even find time to write that much? And that's coming from someone who writes too much on forums!
It doesn't take lot of time when you've been typing since the age of six, programming since the age of eight, and have been programming for a living for some twenty years with a WPM count over 100.
I can type nearly as fast as I think.
dtaylor said:
All I can say is that I only rarely encounter the issues you are complaining about. And I'm usually not using GND filters or a large number of HDR shots, but manually blending two frames.
For a scene requiring HDR I'm guessing that your shadow exposures are not bright enough. I'm also guessing that you are trying to shoot some scenes in one shot when you should have at least two. Your river shot with the blown out sky...I would shoot that as two frames on Canon or Nikon.
I can only guess because I've never been out shooting with you to observe what you're doing. But you talk as if every landscape you do has horrendous shadow noise. If that's happening then you need to adjust your shooting and processing.
You've built up in your mind how much better an Exmor sensor would be, how it would revolutionize your workflow. It's better, but it's not going to revolutionize your workflow or eliminate HDR/GND. That doesn't mean you shouldn't buy a D810 or a Sony A7 series if you want one. Just don't build up Exmor so high in your mind that you buy one and end up complaining on their boards.
This has absolutely nothing to do with technique. My technique is not the problem. Neither is it an exposure problem. If you've actually read anything I've posted on the subject of my recent landscape photos, you would already know that I bracket all my landscape photos. Three or five frames usually. I always bracket. Always have, always will. You never really know if you want it for a landscape, so it's just my standard MO.
I've already blended HDRs for all of my photos, but also as I mentioned before...HDR blends are not perfect either, and they have their fair share of artifacts.
The reason I see noise in the shadows is when you expose to preserve the highlights, you push the rest of the exposure down. This is the opposite of ETTR. This is basically what highlight tone priority does. It's ETTL...you shift the histogram to the left, to pull the highlights back from the right-hand edge of the histogram. The goal, to make sure you have the necessary highlight fidelity, is for your saturation level in the image to top out at around 245 on average (8-bit, it scales to 16-bit). You WANT some headroom above that...you don't want to expose right to 255. That's where you end up clipping one or two color channels, which mucks with your ability to recover highlights with accurate color. To do that, you drop the shadows...you bury more detail in the READ NOISE.
This is where your Photographic DR is unhelpful. Your Photographic DR tells you nothing about the literal, physical capabilities of the hardware. Engineering DR, on the other hand, tells you where that read noise floor is, and how it differs from camera to camera. The 5D III has 33.1e- worth of read noise. The 5D II had 27.8e- RN, the 7D has 8.6e- RN. The D800 has 3e- RN. The difference between the 5D III and 5D II is significant...it's 5.3e-. That is more than the TOTAL read noise of the D800!
I'd happily take 5.3e- RN in Canon's next DSLR. The only reason the 5D III is better than the 5D II in general, and particularly at high ISO, is they bumped Q.E. up from 33% to 49%...THAT is significant, however they castrated themselves at ISO 100 with the huge increase in read noise.
If Canon could release a camera with 5e- RN at ISO 100, and the same FWC as the 5D III, it would have 82.6dB of dynamic range. That comes out to over 13.7 stops of dynamic range. That would solve a LOT of their low ISO IQ problems. However, given Canon's trend...I fully expect RN at low ISO to INCREASE. The 7D had 8.6e-. The 70D has 13.5e- (and with smaller pixels to boot!) The 5D II had 27.8e-, the 5D III has 33.1e-. The 1D IV had 16.6e-, the 1D X has a whopping 38.2e- RN!!!!! (That is a two-fold increase in read noise over the 1D IV...if they had kept the 16.6e- RN with the 1D X, they could have had 12.5 stops of DR.) Canon's current trend demonstrates increases in read noise in each new camera model from the previous generation. I honestly don't know how or why they do that...but, it's the current trend. Maybe the 7D II will change that...but I expect it to end up with something like 15e- RN... ???
My problem with Canon's sensors is a hardware one. I'm forced to make tradeoffs in my exposures, and sometimes I cannot counteract those tradeoffs with things like GND filters. I know how to expose. Of course I do. I know how to bracket and do HDR. IMO, HDR, sky replacement, tonemapping, manual blending, etc. shouldn't be necessary unless you have a truly extreme situation. I don't want to lift shadows to the point where they are midtones. I do, however, want the ability to tweak shadows in a minute or two, and not worry about revealing banding or blotchy color noise or having to increase contrast too much and block up shadows or worse, resort to much more time-costly solutions, so solve the shadow falloff and banding problem. Shadows should remain shadows...but they shouldn't look ugly.
dtaylor said:
So...when it comes to large size images...either something like a 1920x1200 size published online (which I've done a few times for 1x.com...they have a very large format presentation), or larger prints (not sure where the cutoff is, I usually print 13x19), then yes. I HONESTLY do believe that the 5D III suffers from it's shadow noise.
I print a lot at 16x20/24. My albums that I show to family and friends have sleeves for Epson 17x22 sheets so I don't have to cut rolls or trim while filling those. I don't struggle at those sizes...or even larger when I have occasion to print larger. I can literally think of two shots where I did not have a frame with sufficient shadow exposure and was bummed about the noise/tonality/detail in the deep shadows. Of the two, other people have only noticed one.
It's entirely possible I like to shoot scenes with more dynamic range. It's also entirely possible my standards are higher than yours (that's not an insult, people have different standards). Regardless, the shadow noise on Canon cameras requires extra work to eliminate banding, color blotchiness, etc. I'm tired of having to spend extra time fixing things that aren't there on competitor's products. With astrophotography, my time to spend processing is greatly diminished as it is...I have no option but to spend time processing astro images, and the more skilled I get at it, the more advanced my imaging (soon here I'll be moving to a mono camera with color filters, in which case my workload will triple or quadruple, and if I go with both LRGB and NB imaging simultaneously, my workload could compound eight fold...the final results should be FAR superior to what I can do now with a DSLR, but it will require all my time.)
At this point, all I can say is I REALLY hope the 7D II has something Canon's been hiding, like Don says...otherwise I think my loss of confidence in Canon to do anything about their sensor IQ is going to be rather permanent. And, as I said before...that sucks. I don't want to have to buy two different brands, replicate lenses across brands, etc. It's far more cost effective to have a single brand, one set of lenses, and be able to reuse those lenses across bodies. That's why people pick a brand and stick with it in the first place.
A mirrorless high end Canon with EF mount and an EVF! If you think the debates now get rancorous, imagine the hornets nest that would stir up....
Jack
