The GetDPI Photography Forum

Great to see you here. Join our insightful photographic forum today and start tapping into a huge wealth of photographic knowledge. Completing our simple registration process will allow you to gain access to exclusive content, add your own topics and posts, share your work and connect with other members through your own private inbox! And don’t forget to say hi!

16 bit vs 14 bit vs Sony's 11 + 7 compressed file?

mandonbossi

New member
So am really confused and am hoping that someone here may be able to explain to me the difference in bit depth between various cameras/formats.

If I have posted this in the wrong section, please feel free to move it to the appropriate one :)

Here goes... I currently have a Sony A7R and a motley crew of lenses to go with it.. This includes a Canon 24mm TS-E II that I have often used for stitching and have been quite happy with the results.. Resolution wise, once I am looking at a stitched file, it is quite impressive. Many people argue that the resolution of the Sony A7R alone is worthy of mural sized prints..

There are a couple of other options that I have seen for use with the Sony A7R that dramatically increase resolution and that is the Fotodiox Rhinocam

https://www.fotodioxpro.com/vizelex-rhinocam-for-sony-nex-e-mount-cameras.html

and the Hcam Master by Stefan Steib

http://www.getdpi.com/forum/sony/51166-final-version-hcam-master-ts-14-24mm.html


So recently I have seen work by a couple of photographers that I know are using Medium Format Digital and they really have a "look" that is hard to describe.. This goes beyond pure resolution..

The photographers are below;

Ian Allen

Ian Allen Photography

and Magnum Photographer Donovan Wylie

Postcards From America


If you click on either, you should be able to get an enlarged version of the image..

I do remember seeing a link that showed that Donovan Wylie was shooting on a Hasselblad H5D-60 on a 80mm lens..

Cannot seem to find it anymore unfortunately. Am not sure what Ian Allen is shooting on.

So I guess my question relates to color and overall look (CMOS vs CCD) rather than just pure resolution (although this is also a part).

I feel like most people describe the difference between 14 bit (Nikon D810) and 16 Bit (MFDB) it is largely to do with color, whereas when people describe the difference between true 14 bit and Sony's 11 + 7 compressed file, it is more to do with recovering shadows as well as artifacts that occur when doing heavier post processing.. Is there less of a jump bettwen "true" 14 bit and Sony's 11 + 7 Compressed file?

I realize that there may be a lot more going on here with post processing but I would be very grateful to hear any input that you guys have with regards to this..


Thanks a lot! :)
 

Paul2660

Well-known member
I for one don't prescribe to the "look" having shot MFD now since 2007 and film before that (scanned). MFD from either CMOS 50MP or CCD 39, 60 and 80, to me are very close. All files processed out in c1.

As I understand the A7 raw, there is no way to shoot it uncompressed, unless Sony added a firmware update to allow this. So there may be some loss of shadow detail, however in the few shoots I made with the A7r, I did not feel any such loss was there. C1 and LR were used.

The 50MP Sony in Phase One, Pentax and Hasselblad is 14 bit. Phase has 2 levels of compressed raw, just as they do with CCD backs, I always use the larger format. Again, I have never found anything that creates a issue. From the reading I have seen, the 16 bit CCD images have been shown to really only be 14 bit as the last two bits are noise. Where are the Sony @ 14 bit throws nothing out.

In real world processing between a 60MP CCD and 50MP CMOS, I see the following: (all taken with a tech camera and 28, 40 and 60mm lenses no CL-PL used)

1. The 50MP CMOS seems to favor the green channel and so far appears to be a warmer capture than CCD. It's very easy to get an over saturated green color, and you do have to watch for it. Blues seem a bit harder to get where I want them. The CCD 60 and 80 to me have always had a wonderful look to the sky.

2. Shadow recovery is much more possible with the CMOS chip, as much as 2.75 stops at base iso and 2 stops @ 400 iso. This just can't be done with CCD IMO, without excessive noise. Thus bracketing needs to be done.

3. Workflow between CCD and CMOS chips is very different and if start in on a 50MP conversion with the methodology used with a CCD capture, you may run into problems. It's for sure a different raw file. However to my eyes, the end result can be made to look the same.

Paul
 

dougpeterson

Workshop Member
Worry less about bits and tech specs.

Worry more about the final image based on criterium relevant to your needs (do you print? digital only? compositing?).

It's very, very easy to get very deep into the technology/science and end up with LESS of an idea of what is right for you.

EVERY element in the Image Quality Chain is relevant to the technical attributes of the final image. If you start to try to isolate out variables (a reasonable approach in many areas of exploration) what you'll find is that it's rare you can really test for exactly one variable, and not necessarily very relevant given your choices are, for the most part, fixed combinations of components in some specific camera. Here is a partial listing of elements in the Image Quality Chain:
-- Lens coating > Lens elements > Aperture blade design (quantity/shape/size) > internal body coating > microlens (or lack of) > Anti aliasing filter (or not) > IR filter > sensor photo well size/shape > sensor read-out (heat-sinking and/or active cooling very important here) > cables to A/D converter > A/D converter (if not on the sensor) > (read-out of black calibration file from sensor recorded as adjunct to the image) > debayering algorithm > color engine of the raw processor > color profile specific to the camera > deconvolution / detail finding algorithm > noise reduction based on black calibration file > noise reduction based on image data > sharpening.


Go shoot some of these cameras, preferably alongside someone who knows them well and can help you avoid first-time-use sorts of mistakes (a dealer, a friend, a forum member etc), push them through a post-processing workflow that is relevant to you (do you generally print very straight? do you add contrast/dodge/burn?), make some prints (if relevant to you), and see what you think.
 
If we were to isolate bits for a moment, it basically determines how much data you can theoretically cram into a given space, but knowing how large a closet is doesn't say anything about it's contents.

In theory, you only need as many bits of data as the number of stops your camera is capable of capturing to avoid compression, but since most digital cameras cut off at just around 14 stops DR, you never really come close to reaching a point where the DR has to be compressed to fit into a smaller bit depth. I say "in theory" because this is based on the idea that we're dealing with a perfect signal, which anything short of computer graphics is never perfect, even in a 14-bit file the noise floor of most cameras comes in before you run out of room.

This is similar to the discussions sound recordists have with regards to audio equipment. Many claim that 24-bit offers greater flexibility in post-production, but 16-bit still offers 96dB of dynamic range, which is the difference between a whisper on the other side of a room and standing next to a jackhammer. For reference, a perfectly mixed album has on the order of 20~24dB DR and most modern pop songs only around 8~10dB, so those big numbers ultimately need to be compressed to hell and back so that you can actually hear anything on your consumer speakers.
 

Oren Grad

Active member
I feel like most people describe the difference between 14 bit (Nikon D810) and 16 Bit (MFDB) it is largely to do with color, whereas when people describe the difference between true 14 bit and Sony's 11 + 7 compressed file, it is more to do with recovering shadows as well as artifacts that occur when doing heavier post processing.. Is there less of a jump bettwen "true" 14 bit and Sony's 11 + 7 Compressed file?
The main problem with Sony is not bit depth per se, but the lossy data compression algorithm that is used, which can generate artifacts at high-contrast boundaries - not just in "heavier post-processing".

RawDigger: detecting posterization in SONY cRAW/ARW2 files | RawDigger

diglloyd: Sony's lossy file compression
 

Jack

Sr. Administrator
Staff member
My .02,

I doubt you will ever "see" any probative difference between a true 14bit file and a true 16bit file (which may not exist anyway...) 12 and 14, maybe in some super saturated or other boundary colors. 11+7 takes a back seat due to the lossy nature, and you can get visible posterization at both ends when pushing the file around. (Nikon's 14-bit capture for example is far superior to Sony here.)

As for the difference between MFDB capture and full-frame DSLR, I suspect the difference has more to do with the artists eyes than anything else. One notable comment on the two MF shooters you linked to, is both process to the easy side of saturation and leave it in a more natural state, where a vast majority of DSLR shooters will more often pump saturation, sometimes to almost garish levels. There is also something to the general look generated by medium format lenses, they're sharp but generally have a bit lower contrast than the typical smaller format lens, and lower contrast can help render a bit more shadow and highlight detail. Subtle, but difficult to replicate in post since you cannot enhance what wasn't captured to begin with -- and I expect why a number of DSLR shooters favor older legacy lenses... The longer focals for similar framing in MFDB capture also handle DoF fall-off differently. Again subtle, but often visible, and can impact the way an image presents.
 

turtle

New member
Sony compressed files are compromised in theory and in extreme cases, but I have never seen issues in my use despite working files hard. I believe negative effects are extremely rare, or at least only common with specific niche usage that triggers issues.

I am very confident that the look of the photographers you mention is everything to do with what they shoot and how they shoot it than anything else. You can achieve the same look with modest print sizes using a 12 MP M43 camera costing $400. Jack nailed it I think.

Having spent a good deal of time playing with 50MP Pentax 645Z DNGs, while there are benefits over the A7R, they are not as enormous as one might expect and you'll need a large print for it to matter much at all. With aperture and focal length adjusted for format, there is no format look IMO. And if there are issues I am not aware of, I have yet to see them in other people's images.

I'd suggest sticking with the A7R and working out what photographers are doing to make their images look the way they do. It not just about post, its about composition and theme-based editing too.
 
Last edited:

iiiNelson

Well-known member
I think the majority of what you're seeing in the difference between MF and 35mm has more to do with sensor size (and possibly pixel pitch) than the bit depth. As many said you will achieve a MF look even with older MF cameras compared to 35mm.

The gap in IQ has narrowed with more manufacturers being cognizant in lens development with high MP 35mm cameras. MF has done this for some time now as the MP's have been pretty high for some time for them.
 

Shashin

Well-known member
The other difference between the MF and 35mm "look" is that 35mm has extreme focal lengths at both ends and the format aspect ratio. Don't use such extreme optics and change the aspect ratio and you can get "the look."

Any image can be made "mural" size regardless of pixel resolution or format. You just need a big printer.
 

jerome_m

Member
So recently I have seen work by a couple of photographers that I know are using Medium Format Digital and they really have a "look" that is hard to describe.. This goes beyond pure resolution..

The photographers are below;

Ian Allen

Ian Allen Photography

and Magnum Photographer Donovan Wylie

Postcards From America
I am a bit confused because Ian Allen and Donovan Wylie pictures have very different colours and "looks". Could you please post direct links to the kind of pictures you want to take?

I do remember seeing a link that showed that Donovan Wylie was shooting on a Hasselblad H5D-60 on a 80mm lens..
This picture, for example, could have been taken on an Hasselblad camera. The colours are typical of what I get with my H4D and Phocus under winter light: well balanced contrast, with a bit of extra magenta. Is this what you want?

Am not sure what Ian Allen is shooting on.
Some of his pictures appear to have been shot on film. Death valley, Brazil, Hong Kong and Iceland are said to be shot on a large format camera, so film. Ecuador was shot on a Phase back P45+, V mount, the exifs are in the pictures. But Ecuador does not seem to have particular colours to my eyes, others than what is expected at high altitude.
Which of the photographs by Ian Allen were you thinking about?
 
Last edited:

tcdeveau

Well-known member
I've definitely noticed some of the compression artifacts at high contrast boundaries in my own A7R shots. Can't say I'm a big fan of the Sony compressed RAW and really wish they would offer an uncompressed option. My enthusiasm for the system was definitely diminished after closer inspection of files from my first trip with the camera, although I need to actually make some large prints (one of the reasons I went with this system over a Fuji X-T1, at least 24x32) before judging too much. Attached is a screenshot from Lightroom zoomed in 2:1 on a shot from a recent trip to Chile. Granted it's zoomed in and you might not notice this on smaller prints, but artifacts are definitely there (they show up on both my 27" Apple Thunderbolt Display and my NEC PA-272) and may show in larger prints.
 

ErikKaffehr

Well-known member
Hi,

Sony's compression is described here:
RawDigger: detecting posterization in SONY cRAW/ARW2 files | RawDigger


I don't think I have seen this in my own images, but my camera offers different compression levels. From the description, the algorithm should not produce artefacts except on very steep and deep gradients. There is at least one image showing these artefacts clearly:



If you see some other kind of artefacts it is probably not a compression error.

Regarding the MFD look, I shoot Sony and Hasselblad V with P45+. I would say that the P45+ has a clear resolution advantage, but I am pretty sure I cannot observe a difference between A2 size prints. At A1 size I am pretty sure the P45+ wins.

Colour rendition is a bit problematic to compare, it is very hard to process two real world images identically. Shooting a ColorChecker obviously help but I normally don't shoot ColorCheckers.

Here are two images I shot yesterday:



Here the image at top is the P45+ using a Macro PLanar 120/4 at f/16 and bottom is Sony Alpha 99 using a Minolta 100/2.8 macro at f/11.

The raw images are here: http://echophoto.dnsalias.net/ekr/Articles/MFDLook/

Just to say, artefacts can arise for unexpected reasons, this is an out of focus area on a P45+ image:

The very same image is very different with another "camera profile" in Lightroom:


The difference between the two is that a "camera profile" generated by Adobe DNG Profile Editor was used in the first case Adobe Standard profile was used in the second one.

I also tested Capture One, it also produced artefacts with it's internal profile for the P45+.


Here is an article I started writing on the colour rendition issue, it never got published: http://echophoto.dnsalias.net/ekr/Articles/OLS_OnColor/OnColor.pdf

Best regards
Erik
 

turtle

New member
Hi,

I can't see any issue with this. Just looks like a very blurry mountain line against sky. If there are issues here, they aren't obvious.

I've shot thousands of frames with this sort of general daylight scenery and not seen didly squat from the A7 and A7R, even after bending files around like crazy, but I am using a different computer and screen (possibly also software) to you so there are variables.

The star trails, perhaps, but this sort of scenery, I am highly suspicious because I don't think anyone has claimed artefacts under such conditions. Odd things happen with files on screen sometimes, so I'd be amazed if you see anything in print.

We can go crazy looking for problems and end up spoiling things for ourselves. Widespread over-sharpening has created far more problems with the A7R than file compression IMHO.

I've definitely noticed some of the compression artifacts at high contrast boundaries in my own A7R shots. Can't say I'm a big fan of the Sony compressed RAW and really wish they would offer an uncompressed option. My enthusiasm for the system was definitely diminished after closer inspection of files from my first trip with the camera, although I need to actually make some large prints (one of the reasons I went with this system over a Fuji X-T1, at least 24x32) before judging too much. Attached is a screenshot from Lightroom zoomed in 2:1 on a shot from a recent trip to Chile. Granted it's zoomed in and you might not notice this on smaller prints, but artifacts are definitely there (they show up on both my 27" Apple Thunderbolt Display and my NEC PA-272) and may show in larger prints.
 

turtle

New member
tcdeveneau, what are you seeing on the shot you posted, as it may present here differently to what you see on your screen.
 

Paratom

Well-known member
I am also not sure how much and if it has something to do with bit depth or not.
As a user of the Sony A7II, latley an Nikon DF (before a 5dIII as dslr) and Leica S I do however find the images from the Leica S stand out in IQ (and I feel the same when I browse through the internet, even at lower sizes / on screen.) Specially in regards of colors, which seem strong but not oversaturated, and with tones (I think more subtile midtones) and nice sharpness transition.
I also think it has to do a lot with lenses, and with sensor size.
I havent experienced any problems with the Sony files so far, but there is written so much about it that I start to believe it ;)
ANother area where MF seems very good is skin tone. My other cameras in some light produce nice skin tone but in certain situations its difficult (and also difficult to adjust). With the S I feel to get very easy what I want.
 

turtle

New member
maybe this (adjusted and upscaled to make it more obivous) ...
OK, that's clear now. I don't have the expertise to comment on whether that is compression artefacts, or whether they would be visible on print, but I can certainly see the blocks now. Thanks.

The border between the mountain and sky does not appear particularly high contrast and there appears to be no consistency in where they showed up on this image. V odd. I have shot far higher contrast boundaries and not seen anything like this.

Any thoughts anyone?
 

iiiNelson

Well-known member
Looks like image compression to me. I don't know if it's due to Sony's RAW file compression, your RAW processor sharpening, or the PP techniques you used to be honest without checking the original file. Maybe try and see if you see the same thing in Sony IDC.

In any case I'm not sure it would show up clearly visible in prints though at normal print sizes since that's zoomed in way beyond what most people would see. Most large prints are comparable to 33-50% zoom depending on monitor size and the output resolution of a good file for a "reasonable sized print" based on the amount of data you're originally pulling from. For reference I usually print no larger than an inch on the long side per megapixel at 360dpi... but that's just the way I do things though.
 
Last edited:
V

Vivek

Guest
OK, that's clear now. I don't have the expertise to comment on whether that is compression artefacts, or whether they would be visible on print, but I can certainly see the blocks now. Thanks.

The border between the mountain and sky does not appear particularly high contrast and there appears to be no consistency in where they showed up on this image. V odd. I have shot far higher contrast boundaries and not seen anything like this.

Any thoughts anyone?
What Tre said.

(I see blocks all over the frame and not where Thomas circles in red for whatever reason.)
 
Top