Luminous Landscape Forum
Equipment & Techniques => Landscape & Nature Photography => Topic started by: Jonathan Wienke on August 15, 2005, 03:04:38 am
-
Bit depth has no direct relation to color gamut. If you're having color issues with a DSLR it's far more likely a color management problem than an intrinsic limitation of the camera. Most DSLRs can record colors outside Adobe RGB, well outside what's printable in most cases.
-
The size of the gamut is not an issue with the number of bits, but more bits allow you to record finer changes in value within the gamut. Hopefully the device has good enough noise characteristics that the extra bits aren't all used up by noise...
By the way, when Photoshop says it's dealing with 16-bit data it's actually only using 15 bits, but that's still a step up from 4096 steps (12-bit) to 32768 steps (15-bit). In fact because of the maths being used by Photoshop it's not exactly 32768, but close enough for this comparison.
-
The DSLRs all come with 12 bit data. The film scans and the medium format backs (rented not bought) come in 16 bit flavors with an apparent major step up in color gamut. Does this mean anything to anyone or am I being driven crazy?
-
I understand that bit depth is more about noise and dynamic range. 48 bit film scans generally have less noise than 24 bit scans. However, the dynamic range of a sensor has physical limitations depending on the strength of internal noise sources as well as the size of the photosites and the maximum electronic charge they can hold. 16 bit A/D converters and 16 bit signal processing are just trying to make the best of what's there. They can't create something from nothing.
-
12 bits (per channel) colour = 69 billion colours
16 bits (per channel) colour = 28 trillion colours
A gamut is a collection of shades nothing more. A ggamut could contain a gazillion colours but 12 bit file could only contain 69 billion of those shades simultaneously. Where as the 16 bit can contain 28 trillion colours. So yes 16 bits per channel photo should look better
The DSLRs all come with 12 bit data. The film scans and the medium format backs (rented not bought) come in 16 bit flavors with an apparent major step up in color gamut. Does this mean anything to anyone or am I being driven crazy?
[a href=\"index.php?act=findpost&pid=3065\"][{POST_SNAPBACK}][/a]
-
I understand that bit depth is more about noise and dynamic range. 48 bit film scans generally have less noise than 24 bit scans.
I don't think so. First, a lot of this is theoretical math. 12 bits 69 billion colors, 16-bit 28 gizillion. Use either capture device and shoot a gray card and how many colors are you getting?
As for noise, I don't see how more levels equate to less or more noise OR dynamic range. You're just splitting the possible numeric values captured into finer numbers. How would that equate to less noise or more range? These would appear to be totally different specs based on a lot of factors.
In my mind, high bit (more than 8-bit) is about editing overhead. I want to send the best 8-bits to an output device. If I start with 256 values and edit the image, the net results are less than 256 values. If the same data starts with 64000 values, I can edit and end up with more than enough levels to send the best 256 to the device. Doesn't mean that provides better appearing images, less noise etc, it just means I start with finer steps of numeric values.
-
Andy you may well be right as 16 billion cols is large no., it's not exactly like viewing a photo in like say 16 cols. I'm not sure how many colours the average human can see?
-
Gamut is not really the number of different colours a system can see, but more how different the different colours are. 12bit / 16bit governs dynamic range / noise floor and number of different tones. Gamut is defined by the colour filters on the sensor, image processing etc.
And when a camera claims 16bit, you've got to go and look at the dynamic range and noise floor, as some of those bits couldbe just pure noise, or even blank.
Graeme
-
Opening no old debates here - promise - but how simple it was in the days of film! You had more or less grain, or your film had dyes and practically no grain. Now some of us have headaches instead of confidence.
Yes, that´s progress, folks, modern style!
Ciao - Rob C
-
Just because an analog to digital converter is a 16 bit device there is no guaranty that the least significant bits aren’t lost in noise and distortion. In audio 24 bit a/d converters you’re lucky if the 14 most significant bits are valid data. I'm curious what the usable bits are in both 12 bit DSLRs and 16 bit digital backs? I’m sure less than the a/d converters bit depth as they don’t have DAC ovens, controlled impedance circuitry etc.
Marc
-
How bit depth was explained to me.....
Think of bit depth as a ladder....as you increase bit depth...you dont extend the size of the ladder...just how many rungs are between each end.
Roman
-
All you people worry about bits, do you really see it? in a print? Everyone talking about DSLR at 12 bit. Look again the Canon 1D Mk III is 14 bit. It might be the first Canon at 14 bit but it won't be the last. So, are you willing to paid over $20,000 grand for a MF to get the 16 bit.
I shot the Horseshoe bend at Page, Az. Both with a 1Ds MkII and Phase One P45, looking at both images, there is little difference in the images. Yes the P45 does show more, but at what price? 1Ds MkII at $8000 or the P45 at $33500, that is just the camera body and MF back.
-
All you people worry about bits, do you really see it? in a print? Everyone talking about DSLR at 12 bit. Look again the Canon 1D Mk III is 14 bit. It might be the first Canon at 14 bit but it won't be the last. So, are you willing to paid over $20,000 grand for a MF to get the 16 bit.
I shot the Horseshoe bend at Page, Az. Both with a 1Ds MkII and Phase One P45, looking at both images, there is little difference in the images. Yes the P45 does show more, but at what price? 1Ds MkII at $8000 or the P45 at $33500, that is just the camera body and MF back.
[a href=\"index.php?act=findpost&pid=132186\"][{POST_SNAPBACK}][/a]
Would you post both images as a comparison? (I know it is hard on the web to compare) I'd be curious, not concerned about bits but real world dynamic range. Initial reports of the Mamiya zd back show it to be a noticeable improvement over the 5D in this regard.
Marc
-
I guess everyone is waiting for me, thanks to Marc, I will try to post the pics. The 4:3 is the P45 shot with the Mamiya 35mm at f8, the 3:2 was shot with the Canon 1Ds MkII with the 24-70 zoom at 24mm and f8, both was output with C1. They are full frame and nothing was done to the images, except to make them jpegs. The color and contrast does not match, because nothing was done to them. They are taken about 1 minute apart.
-
http://www.mobypicture.com/user/littletonhairsalon/view/17414018
http://www.last.fm/user/littletonsalon
http://www.kiva.org/lender/botox
http://denver-colorado.weebly.com/
http://littleton-juvederm.weebly.com/
http://plancast.com/p/lrw9
http://www.salonandspa.services/component/tags/tag/17-juvederm-in-littleton
http://www.salonandspa.services/our-services/juvederm-littleton
http://botox.flavors.me/
-
The DSLRs all come with 12 bit data. The film scans and the medium format backs (rented not bought) come in 16 bit flavors with an apparent major step up in color gamut. Does this mean anything to anyone or am I being driven crazy?
[a href=\"index.php?act=findpost&pid=3065\"][{POST_SNAPBACK}][/a]
You can't readily tell why the MF backs are better though. It can be for other reasons besides bit depth. Just because a manufacturer uses high bit depth, doesn't mean that it is actually used in any truly beneficial way by the camera, although it may force a converter to use a higher precision; it may simply "pad" the least significant bits with noise instead of zeros.
-
12 bits (per channel) colour = 69 billion colours
16 bits (per channel) colour = 28 trillion colours
A gamut is a collection of shades nothing more. A ggamut could contain a gazillion colours but 12 bit file could only contain 69 billion of those shades simultaneously. Where as the 16 bit can contain 28 trillion colours. So yes 16 bits per channel photo should look better
[a href=\"index.php?act=findpost&pid=128971\"][{POST_SNAPBACK}][/a]
In reality, noise is humongous compared to digitized RAW levels in the highlights, and significant in the shadow areas as well. Bit depth is not a limit in most digital cameras; most have more bits than can be efficiently utilized. Extra bits in the RAW files guarantee higher conversion precision at best, in most cases, but that can be done by just fabricating the LSBs.
-
With a Stouffer 13 1/3 stop test wedge, I could clearly see more steps (about 2 stops worth) on a Canon 1D MK III than a Canon 20D.
Graeme
-
I don't think so. First, a lot of this is theoretical math. 12 bits 69 billion colors, 16-bit 28 gizillion. Use either capture device and shoot a gray card and how many colors are you getting?
As for noise, I don't see how more levels equate to less or more noise OR dynamic range. You're just splitting the possible numeric values captured into finer numbers. How would that equate to less noise or more range? These would appear to be totally different specs based on a lot of factors.
In my mind, high bit (more than 8-bit) is about editing overhead. I want to send the best 8-bits to an output device. If I start with 256 values and edit the image, the net results are less than 256 values. If the same data starts with 64000 values, I can edit and end up with more than enough levels to send the best 256 to the device. Doesn't mean that provides better appearing images, less noise etc, it just means I start with finer steps of numeric values.
[a href=\"index.php?act=findpost&pid=128991\"][{POST_SNAPBACK}][/a]
Andrew,
I'd forgotten I had expressed an opinion on this matter (2 years ago!!).
It's not clear to me precisely under what circumstances 16 bit will be noticeably advantageous compared with 8 bit. However, if we start from your premise that 16 bit is all about editing headroom, then let's ask the questions, 'When are images never edited? Have I ever seen a completley unedited image? What does it look like?'
There's such a thing as in-camera processing, even of RAW images and certainly of jpegs. That's editing, isn't it?
When I scan my slides, I make all sorts of adjustments to the preview of the slide before I hit the 'scan' button. That's editing isn't it?
When I shoot a high contrast scene with my Canon DSLR, there will be a very limited number of levels available to describe the deepest shadows in 8 bit mode, more in 12 bit and even more in 16 bit. If I'm trying to get the maximum DR and the least shadow noise through all sorts of in-camera processing (which would appear to be a trade secret since I've never seen such processes explained in detail), then it would be reasonable to suppose that 16 bits per channel are better than 8 or 12 bits, wouldn't you say?
-
With a Stouffer 13 1/3 stop test wedge, I could clearly see more steps (about 2 stops worth) on a Canon 1D MK III than a Canon 20D.
[a href=\"index.php?act=findpost&pid=132705\"][{POST_SNAPBACK}][/a]
But why? You'd see them on a mk2, also, because the read noise is the same, relative to max signal, at ISO 100. The pixel read noise is lower in the mk3 than the 20D at all ISOs, plus it has more pixels, so if you achieve the same FOV, you will have even less subject read noise.
I've quantized ISO 100 RAWs from the mk3 to 12 bits, and they are indistinguishable from the full 14 bits after RGB interpolation, even zooming deep into the shadows, as long as the 12 bit version is promoted back to 14 bits with zeros in the 2 LSBs.
So, more bits might help for indirect reasons, and not necessarily because of what they contain. In my experience, working bit depth in conversion is generally more important than the significant bit depth of the RAW data. That's because there is so much noise, and the noise gets softened in conversion, and the greatest dangers of posterization occur in the rendered product.
-
Andrew,
I'd forgotten I had expressed an opinion on this matter (2 years ago!!).
It's not clear to me precisely under what circumstances 16 bit will be noticeably advantageous compared with 8 bit. However, if we start from your premise that 16 bit is all about editing headroom, then let's ask the questions, 'When are images never edited? Have I ever seen a completley unedited image? What does it look like?'
There's such a thing as in-camera processing, even of RAW images and certainly of jpegs. That's editing, isn't it?
I'm teaching all week with limited time to follow this discussion. I will say no, Raw processing is not editing, its rendering. There's no data loss, its true non destructive since the pixels are rendered from the Raw data (they are forgive the term, virgin pixel data) rendered from the high bit Raw data. Rendering is a bit different from editing of existing pixel numeric values.
-
I'm teaching all week with limited time to follow this discussion. I will say no, Raw processing is not editing, its rendering. There's no data loss, its true non destructive since the pixels are rendered from the Raw data (they are forgive the term, virgin pixel data) rendered from the high bit Raw data. Rendering is a bit different from editing of existing pixel numeric values.
[a href=\"index.php?act=findpost&pid=132808\"][{POST_SNAPBACK}][/a]
Well, I suppose we won't hear from you for a few days.
I understand that the conversion or rendering of the RAW data to a viewable image format is not considered to be editing. The unanswered question here is, what processing, if any, takes place after the A/D conversion but prior to the recording of the RAW data to the memory card?
We know that in Nikon cameras such as the D2X there's an option to turn on noise reduction, which unfortunately reduces resolution. In the 5D there appears to be some noise reduction taking place at high ISO's because image resolution suffers (in RAW images) at ISO 1600 and 3200, more so than in the 1Ds2.
It appears that 12 bits are sufficient to record the full dynamic range of a Canon DSLR. Although the pixel pitch on the sensors in MF backs such as the P45 is no greater than that of some Canon DSLRs, in fact slightly smaller at 6.8 microns than the pixel pitch of the 1Ds2 at 7.2 microns, 16 bits are apparently needed to accommodate the wider dynamic range of the typical MFDB, presumably because of the greater fill factor and well capacity of the CCD which doesn't have to make room for on-chip processors.
BTW, the first post in this thread appears to begin with Jonathan Wienke's reply to the OP. What happened to the original post I wonder?
-
I've quantized ISO 100 RAWs from the mk3 to 12 bits, and they are indistinguishable from the full 14 bits after RGB interpolation, even zooming deep into the shadows, as long as the 12 bit version is promoted back to 14 bits with zeros in the 2 LSBs.
[a href=\"index.php?act=findpost&pid=132740\"][{POST_SNAPBACK}][/a]
John,
Why did you not also quantize ISO 6400 RAWs from the 1D3 to 12 bits?
-
I thought the Fuji and new canon are true 14-bit not 12-bit?
[a href=\"index.php?act=findpost&pid=132647\"][{POST_SNAPBACK}][/a]
Canon's release on the new 1DsMkIII says it is 14 bit.
-
John,
Why did you not also quantize ISO 6400 RAWs from the 1D3 to 12 bits?
[a href=\"index.php?act=findpost&pid=133081\"][{POST_SNAPBACK}][/a]
Sorry, I didn't notice your reply until today. This forum is a little hard to keep up with with my approach (to look at "View New Posts"), as the high level of activity on this site makes things scroll out of the first three pages pretty quickly sometimes.
I didn't quantize ISO 6400 RAWs because I don't have any, and because I know that ISO 100 or 50 are the most likely to benefit from extra bits. I don't think that ISO 1600 is even worthy of 12 bits; 11 would be sufficient. There is a lot of waste in these cameras, with the manufacturers trying to impress us with bit depths that are overkill with the current noise levels.
-
I don't think that ISO 1600 is even worthy of 12 bits; 11 would be sufficient. There is a lot of waste in these cameras, with the manufacturers trying to impress us with bit depths that are overkill with the current noise levels.
[a href=\"index.php?act=findpost&pid=137574\"][{POST_SNAPBACK}][/a]
I'm struggling to justify upgrading from my 20D to the 40D. I know there are some nice features such as a faster frame rate, ISO bracketing, faster focussing etc but it's difficult to see any fundamental performance increase in either resolution or noise.
I'm wondering how useful that faster frame rate would be for hand-held bracketing for HDR purposes, now that CS3 has a much improved auto-alignment feature.
This indecision is compounded by Michael's claim, in his review of the 40D, that image quality is on a par with that of the 5D, at least as good. If that's true, then my 5D would become redundant. I already own an EF-S 10-22 zoom, which is roughly equivalent to my Sigma 15-30 on the 5D. With just 3 lenses and a 1.4x extender I'd have an FL range from 16mm to almost 900mm. (10-22mm, 24-105mm, 100-400mm).
If only Canon had given us a bit more with the 40D, 12mp instead of 10mp, an additional ISO of 6400 like the 1D3, then I would not have such hesitation.
I think I might wait for the 5D Mk II.
-
I'm struggling to justify upgrading from my 20D to the 40D. I know there are some nice features such as a faster frame rate, ISO bracketing, faster focussing etc but it's difficult to see any fundamental performance increase in either resolution or noise.
I have only seen ISO 100 RAWs so far, but they have about 1/2 stop more DR (read noise drops from 2.07 to 1.38 ADU (adjusted to 12-bit).
Comments I've read from people who already have 40Ds are that the shutter is quieter, and the camera is more responsive all around.
I'm wondering how useful that faster frame rate would be for hand-held bracketing for HDR purposes, now that CS3 has a much improved auto-alignment feature.
This indecision is compounded by Michael's claim, in his review of the 40D, that image quality is on a par with that of the 5D, at least as good.
That depends what he means by that; the FF has the advantage of higher line resolution per frame from the same lens; this and the fact of the higher pixel count were the only reasons the 5D gave better images than the 20D. Electrically, each pixel on the 5D is almost identical to the 20D from ISO 100 to 1600, except the 5D had a slight disadvantage in DR, due to a lower highlight clipping level.
If that's true, then my 5D would become redundant. I already own an EF-S 10-22 zoom, which is roughly equivalent to my Sigma 15-30 on the 5D.
[a href=\"index.php?act=findpost&pid=137593\"][{POST_SNAPBACK}][/a]
The sigma on the FF is significantly sharper than the 10-22 I would think. My 15-30 is very sharp, but its downfall is its "sunflare".
-
The sigma on the FF is significantly sharper than the 10-22 I would think. My 15-30 is very sharp, but its downfall is its "sunflare".
[a href=\"index.php?act=findpost&pid=137801\"][{POST_SNAPBACK}][/a]
I've just compared the 10-22 on my 20D at 10mm with the Sigma 15-30 on my 5D at 15mm. That 1mm difference in effective focal length seems surprisingly significant. I had to move the tripod back to get the same FoV as the 5D.
There's no doubt that the 5D/Sigma 15-30 combination produces sharper results than the 20D/10-22, at the shortest FL. Without further testing, it's difficult to say to what extent this is due to softness in the 10-22 lens and to what extent it's due to the fewer pixels of the 20D. The main point is, I think it unlikely that the 40D with 10-22 zoom would be on a par with the 5D/15-30.
-
Hmmm. I want to pick up on the discussion a few comments back, about bit depth and noise and colour gamut, before it got on to lenses.
I would guess that many of the topics being discussed here are independent variables, and are being mixed together into a bit of a messy soup. Noise, for example, is generated by each photosite, along with the signal. Every electronic device creates some noise. If the signal level is very low (dim light) then noise becomes a bigger issue, starting to overpower the signal itself. Hotter sites also generate more noise and there has been discussion about the heating of various sensors and how that affects the image. Amplification of the signal also amplifies noise, so higher ISO (higher amplification) means more noise. If a photosite could be built that generates no noise (impossible) then there would be no noise, irrespective of bit depth, 2 or 22. Noise and bit depth are independent variables.
Bit depth is a design feature of AtoD converters. Converters are electronic devices that are built with a specific bit depth. That is just a physical feature of the converter. An analogue signal being converted to digital will have more detail saved if converted with an AtoD converter that has more bits. Audio fanatics talked about this a lot when CDs were replacing LPs.
So, not having looked into the details, I am guessing that the analogue signal received through the lens (this device does not count photons per se) is converted to a digital signal after each photosite. I’m sure there are not 16 million A to D converters on the chip(s), so it must be somehow streamed sequentially to a number of parallel converters. Whoever talked about detail between limits is right. The A to D converters can only use the dynamic range fed to them by the photosites. Again, these are two independent operations. The bit depth does not extend or reduce the dynamic range (unless the Ato D converter was very badly designed, which I doubt). The AtoD converter simply stores more or less detail within the photosite’s dynamic range, and, I’m sure, is designed to cover the total range.
Then there is colour. Each photosite is sensitive to a range of the electromagnetic spectrum and produces a signal when that part of the spectrum ‘excites’ it. This is nonlinear, and I have no idea if the camera manufacturers try to do something to account for that. I could get into colour theory ( but would have to look it up again) so I will just remind that there are 3 sensors per site, one for each of the primary colours. So a measure of the real colour is based on the relative values from each of the three sites. Bit depth should play into this I guess, because the greater the detail about each of the primary colours ( more bits) the greater the accuracy of the final mix.
The last bit is the construction of the sensor. There are lenses at each site. The bigger the lens the higher the signal so the lower the noise (relatively speaking). The lenses, I am again guessing, also have colour filters on them ?? I haven’t really looked into that but assume it is the case, so they can differentiate between the primary colours. Then there are UV filters built in because the photosites do not have the same colour sensitivity as our eyes and we do want the pictures to look the same as when we see the real thing. I haven’t explored the details of sensor spectral sensitivity at all, but the point I want to make is that the colour gamut is a function of the sensor and the filters that are used, not a function of bit depth. It is the fine gradations within the gamut that are a function of bit depth.
If there are any Electronics Engineers/camera/chip designers out there that find fault with my reasoning, please let me know.
-
Well, I think noise, most of which I'm assuming (there's that word) originates in the sensor complex, would be more faithfully reproduced coming from a 14-bit A/D converter ... maybe more aesthetically pleasing?
-
Bit depth is a design feature of AtoD converters. Converters are electronic devices that are built with a specific bit depth. That is just a physical feature of the converter. An analogue signal being converted to digital will have more detail saved if converted with an AtoD converter that has more bits. Audio fanatics talked about this a lot when CDs were replacing LPs.
Let's look at the audio analogy. My enthusiasm for hi fi matters extended from a few years prior to the introduction of the CD to a few years after. As I understand, the bit depth of 16 per channel was a minimum necessary to faithfully reproduce the entire gamut of audible frequencies (20hz to 20kHz) with inaudible noise. I think S/N was typically 98db, a huge improvement on the LP vinyl record which was around 58db without the later compression technology like dbx which was beginning to become popular about the time the CD became available.
The early CD sound was much criticised for its harshness, the explanation being that the sound engineers continued using the same old techniques they'd been used to using with the analog systems which had their own built-in roll-offs and softening effects. I believe there was also a certain harshness introduced by the steep 'brick-wall' filters used in the digital systems. These initial problems have been fixed by oversampling and greater bit depth. I believe DVD Audio is now 24 bit but I think the problems had been fixed without resorting to the greater bit depth. The differences between a well-produced 16 bit audio CD and a 24 bit DVD Audio disc are very subtle as I understand, probably too subtle for my hearing, and certainly require top-end equipment for playback.
I suspect that moving from 12 bit to 14 or 16 bit A/D conversion with relatively small-photosite DSLRs like Canon's is like upgrading from 16 bit audio to 24 bit audio. One struggles to find an improvement.
-
The DSLRs all come with 12 bit data.
The Leica Digital-Module-R is 16-bit.
-
Basically, quantisation noise is 6db per bit, so 16bits has a 96db noise floor, assuming the rest of the circuitry doesn't eliminate that.
Frequency is a function of sample frequency, not bit depth. Frequency must be at least twice wanted maximum recordable frequency to avoid, yes, you guessed it, aliasing. It's a sampled system! Brick wall filters can be built electronically, so you can get very close to the maximum with a very steep filter - something you can't do optically with an OLPF, but in both cases they're necessary. Audio aliasing sounds horrible, just as visual aliasing is ugly.
The main problem, as I see it, with early CD players was that they didn't fully understand all the nuances of digital playback - and they shared a single a to d over two channels, many were only 14 bit. You can take a well mastered CD from way back when and play it on a modern player and it can sound much improved.
That said, a lot of early CDs were mastered by people who didn't understand digital technology or it's nuances. They were used to mastering for vinyl where you'd be able to "get away" with more treble nasties. Remember LPs go through a brutal equalisation curve that CDs just don't have / need.
Let's look at the audio analogy. My enthusiasm for hi fi matters extended from a few years prior to the introduction of the CD to a few years after. As I understand, the bit depth of 16 per channel was a minimum necessary to faithfully reproduce the entire gamut of audible frequencies (20hz to 20kHz) with inaudible noise. I think S/N was typically 98db, a huge improvement on the LP vinyl record which was around 58db without the later compression technology like dbx which was beginning to become popular about the time the CD became available.
The early CD sound was much criticised for its harshness, the explanation being that the sound engineers continued using the same old techniques they'd been used to using with the analog systems which had their own built-in roll-offs and softening effects. I believe there was also a certain harshness introduced by the steep 'brick-wall' filters used in the digital systems. These initial problems have been fixed by oversampling and greater bit depth. I believe DVD Audio is now 24 bit but I think the problems had been fixed without resorting to the greater bit depth. The differences between a well-produced 16 bit audio CD and a 24 bit DVD Audio disc are very subtle as I understand, probably too subtle for my hearing, and certainly require top-end equipment for playback.
I suspect that moving from 12 bit to 14 or 16 bit A/D conversion with relatively small-photosite DSLRs like Canon's is like upgrading from 16 bit audio to 24 bit audio. One struggles to find an improvement.
24bit sound can be subtle - but - just think what a difference it makes to what the mastering engineer can do - it's what I call "production headroom". Remember, we might shoot with a 12 or 14 bit a to d, but we often only see an 8bit representation (ignoring what that one is gamma, one is linear for now). That's our production headroom.
But a 196hz 24bit copy of the mastertape can sound fantastic - with a listen-through clarity and lack of listener fatigue, even with all that extra detail, that CD just doesn't deliver. That's what I'm listening to as I write this forum post, coincidentally.... :-)
Now, when I shot a stouffer test wedge with the MKIII and it's 14bit a to do, it did look to me that there was a couple stop improvement in dynamic range over that of the 20D. I didn't do it fully scientific like I do when I have time, but the sensor did seem to be lower noise, and hence could properly benefit from a better depth a to d.
Greater a to d depth matched by lower noise in the sensor means more dynamic range. That's both easily visible and quantifyable.
As for 16bit backs - without published or tested SNR ratios for the sensor, they're just marketing fluff. Let's see a stepwedge. I don't think their sensor gets close to 96db SNR, but a wedge test would prove me wrong. But you'd need to take two exposures to cover that range as popular test wedges only have a 4.0OD range - or 13.3 stops - not th 16 you'd need to measure properly a 16bit sensor!
Graeme
[a href=\"index.php?act=findpost&pid=144768\"][{POST_SNAPBACK}][/a]
-
You can take a well mastered CD from way back when and play it on a modern player and it can sound much improved.
Especially if you coat the outer edge with a green marker - I've been thinking of treating my sensor like that to lower the noise and smooth out the gradients ...
-
Especially if you coat the outer edge with a green marker - I've been thinking of treating my sensor like that to lower the noise and smooth out the gradients ...
[a href=\"index.php?act=findpost&pid=147605\"][{POST_SNAPBACK}][/a]
Yes!!!
There was an excellent product targeted for cds with the precisely correct shade of green (complementary color of the red laser-beam), now there might be similar market for painting the high-end camera sensors. You can clearly see (hear) the difference (at least in your wallet)...
Cheers,
J
-
Bit depth has no direct relation to color gamut. If you're having color issues with a DSLR it's far more likely a color management problem than an intrinsic limitation of the camera. Most DSLRs can record colors outside Adobe RGB, well outside what's printable in most cases.
[a href=\"index.php?act=findpost&pid=3063\"][{POST_SNAPBACK}][/a]
Well, this might be true, or not.
As far as I know, there are no widespread tests that measure the ability of DSLRs to capture colors accurately at very high or very low illuminations, and most tests focus anyway on the ability to reproduce accurately a narrow range of standard color patches.
1. high illuminations
We know that our DSLR (seen as a black box) are not that great at dealing with subtle differences in illumination near their saturation point. I believe that this impacts their ability to distinguish subtle colors as well.
Whether this is the result of sensor limitations, D/A conversions limitations (including discrete sampling) I am not sure, but I would not exclude the possibility that bit depth (seen as an arbitrary metrics to represent the quality of D/A conversion) impacts the quality of color in high illuminations.
2. subtle color differences (color resolution)
Now, one topic that should also be discussed a lot more is color quality. Why are some DSLRs (Fuji comes to mind) seen as producing more pleasing colors than others?
In the end the outcome is the same, a RAW file file some values that - within a given color space - translate into colors on the screen or on paper.
Is Fuji only playing with the values after capture, of does their sensor really capture color with more quality (resolution) that say Canon or Nikon? Nobody knows for sure, but my (mostly unbustantiated) belief is that their sensor might be able to capture both more colors (wider gammut), but also to differentiate more subtle color differences. This is another area where there might be a link with bit depth since it would take enough bit depth to differentiate the hues captured by the sensor.
Cheers,
Bernard
-
Well, this might be true, or not.
As far as I know, there are no widespread tests that measure the ability of DSLRs to capture colors accurately at very high or very low illuminations, and most tests focus anyway on the ability to reproduce accurately a narrow range of standard color patches.
1. high illuminations
We know that our DSLR (seen as a black box) are not that great at dealing with subtle differences in illumination near their saturation point. I believe that this impacts their ability to distinguish subtle colors as well.
Whether this is the result of sensor limitations, D/A conversions limitations (including discrete sampling) I am not sure, but I would not exclude the possibility that bit depth (seen as an arbitrary metrics to represent the quality of D/A conversion) impacts the quality of color in high illuminations.
2. subtle color differences (color resolution)
Now, one topic that should also be discussed a lot more is color quality. Why are some DSLRs (Fuji comes to mind) seen as producing more pleasing colors than others?
In the end the outcome is the same, a RAW file file some values that - within a given color space - translate into colors on the screen or on paper.
Is Fuji only playing with the values after capture, of does their sensor really capture color with more quality (resolution) that say Canon or Nikon? Nobody knows for sure, but my (mostly unbustantiated) belief is that their sensor might be able to capture both more colors (wider gammut), but also to differentiate more subtle color differences. This is another area where there might be a link with bit depth since it would take enough bit depth to differentiate the hues captured by the sensor.
Cheers,
Bernard
[a href=\"index.php?act=findpost&pid=147711\"][{POST_SNAPBACK}][/a]
What an awful lot of discussion (and I hope that's all of it!) about this part of the whole process in bringing an image to presentation. In my little corner, I think I'm looking at all that really matters when I see that an uncompressed TIF, saved from an edited RAW, normally has at least 40% higher color count than a lossless JPG. I don't have any software that can output and print an edited 16 bit TIF, unless I want to accept the limits of Canon Photo Professional.
My point is that unless you spend all that money for a system that can output a finished 16 bit image to the printer, this discussion is almost entirely moot. Most of us live and print in an 8 bit world, printing somewhere between 600,000 to 1M colors. The whole discussion of perceived gradients is generally answered by which colors fill that spectrum, from the larger number that were available, before the reduction to 8 bit. It's generally safe to say that we're seeing one million or less colors in almost all print media. Beyond that, the brain literally sees what it wants to see, and we would still see the very same thing, whether the color count was 1M or 500B.
If you want to test everything that the discussion has covered, you would have to use a 64 bit computer, running very carefully selected software, and printed by the Epson R1900, with 18.4 trillion color output. It shouldn't be that hard to have fun with imaging!
-
My point is that unless you spend all that money for a system that can output a finished 16 bit image to the printer, this discussion is almost entirely moot.
More bits generally means more headroom for processing.
-
If you want to test everything that the discussion has covered, you would have to use a 64 bit computer, running very carefully selected software, and printed by the Epson R1900, with 18.4 trillion color output. It shouldn't be that hard to have fun with imaging!
Panoak, what you say is basically true, but keep in mind that lots algorithms are applied to the image and it is there where the 16-bits of info are valuable as you transform this and interpolate that. The end 8-bits outputted will be more accurate if the original data was in 16-bits.
But yes, imaging needs to be fun and enjoyable, otherwise there's no point .
-
More bits generally means more headroom for processing.
[a href=\"index.php?act=findpost&pid=147728\"][{POST_SNAPBACK}][/a]
But it's still the same thing. It doesn't matter if you started with 16 million or 96 billion colors. Once you land back in the world of 8 bit, your 600,000 to 1M colors are still all you have, and although the starting number will influence the colors within that range, the result comes down to a difference that is academic, at best.
The really aggravating thing with all of this is that it's very hard to find equipment and software that enables you to put the higher numbers to the test. This stuff is all made to satisfy most of the people most of the time, and it does.
-
1. high illuminations
We know that our DSLR (seen as a black box) are not that great at dealing with subtle differences in illumination near their saturation point. I believe that this impacts their ability to distinguish subtle colors as well.
[a href=\"index.php?act=findpost&pid=147711\"][{POST_SNAPBACK}][/a]
I don't agree with that at all. There's a vast over-abundance of linear code values. In a 12bit system, the brightest highlight stop has 2048 code values, (that's 8 times as many code values in a whole 8bit image, and we're talking about just the brightest stop of highlights in the 2048.)
Graeme
-
We know that our DSLR (seen as a black box) are not that great at dealing with subtle differences in illumination near their saturation point. I believe that this impacts their ability to distinguish subtle colors as well.
That's only true with some cameras, at the lowest ISO. The sensor can be non-linear near saturation, or the the camera may stretch the highlights to make it appear that the full range of RAW values are used (such as the Canon 20D, which should really have had a base ISO of about 115). Even then, there is a potential for a RAW converter to have specific information about a camera, by running a series of tests, to see how linear it is, where clipping occurs, detect fixed pattern noises, etc, to give optimal conversions from that camera. I've never heard of any converter offering this possibility, though.
Is Fuji only playing with the values after capture, of does their sensor really capture color with more quality (resolution) that say Canon or Nikon? Nobody knows for sure, but my (mostly unbustantiated) belief is that their sensor might be able to capture both more colors (wider gammut), but also to differentiate more subtle color differences. This is another area where there might be a link with bit depth since it would take enough bit depth to differentiate the hues captured by the sensor.
[a href=\"index.php?act=findpost&pid=147711\"][{POST_SNAPBACK}][/a]
More bit depth would help, but it only helps when noise is at bay. Noise is far more significant than bit depth in most current cameras. Quantization of RAW data is not an issue with current cameras. You see quantization in the output for two reasons:
1) An optical illusion, where your mind groups similar colors of individual pixel noises as similar - imagine a canvas of grey 128, and a dozen isolated squares of blue, all with slightly different values - do they look any different to you? You may have to use the info tool, or exaggerate the differences with the levels tool to tell them apart.
2) Converters don't use enough bit depth - no room for extended precision in the deepest shadow areas, and noise-softening applied too agressively in the highlight areas. A tiny amount of noise is always good, when your display medium is posterized.
-
But it's still the same thing. It doesn't matter if you started with 16 million or 96 billion colors. Once you land back in the world of 8 bit, your 600,000 to 1M colors are still all you have, and although the starting number will influence the colors within that range, the result comes down to a difference that is academic, at best.
I find that original 16-bit data holds up better when I need to stretch the histogram, giving much better gradation than if I were to try the same thing with lower bit depth.
-
16bit data does, but it also does so almost independent of the dynamic range that generated that 16bit data.
Graeme