December 03, 2016, 07:14:19 AM

Author Topic: More Big Megapixel Talk [CR1]  (Read 85060 times)

WoodyWindy

  • Rebel SL1
  • ***
  • Posts: 79
  • On the road again!
Re: More Big Megapixel Talk [CR1]
« Reply #90 on: September 26, 2012, 12:55:28 PM »
Between this and the rumor about a revamp of the APS C line, I'd be guessing that:
1. There is a new sensor technology coming that will be used across these units
2. The pixel pitch isn't changing much
3. That per-pixel exposure calculation patent sounds interesting
4. Combined with rear illumination (which, I'll say again, I'm shocked that it is even considered an innovation. I'd always assumed that was the way sensors worked...)
5. Combined with pixel binning of some form.
6. Probably a new series of Digic to process the mess.

What about a depth sensitive sensor, sorta like Foveon, but eliminating the red issue with a new color filter array, with alternating pixels sensing (by depth) two colors each, RG and BG. Leave the BG pixels unfiltered so you can get full LUM data, Green depth, and blue depth. Then have a yellow filter over the alternating pixels to let red and green have at it. You get 100% green coverage, and 50% each true b&w, red, and blue, instead of 50% green, 25% red, and 25% blue, and no full luminance data. If you wanted to get real fancy, you could still sense for a noisier red value from the unfiltered pixels, giving you 100% red coverage in good light.

canon rumors FORUM

Re: More Big Megapixel Talk [CR1]
« Reply #90 on: September 26, 2012, 12:55:28 PM »

HurtinMinorKey

  • Canon 6D
  • *****
  • Posts: 446
    • carolineculler.com
Re: More Big Megapixel Talk [CR1]
« Reply #91 on: September 26, 2012, 12:56:16 PM »
The reason why it isn't offered is because it takes extra processing to make a med raw file meaning way longer than a human can tak for a file size that large unless it had CFast 2 the new compact flash card system and stellar buffer.

Sorry, I don't understand your point about extra processing time.  Canons shoot just as fast in mRAW and sRAW as in full RAW.

Semantic arguments aside about what does and does not constitute compression, using mRaw is just a different read off the sensor. It doesn't require any extra processing.  Raw data is maintained, albeit from fewer photosites.

jrista

  • Canon EF 600mm f/4L IS II
  • **********
  • Posts: 5334
  • EOL
    • Nature Photography
Re: More Big Megapixel Talk [CR1]
« Reply #92 on: September 26, 2012, 01:28:11 PM »
The reason why it isn't offered is because it takes extra processing to make a med raw file meaning way longer than a human can tak for a file size that large unless it had CFast 2 the new compact flash card system and stellar buffer.

Sorry, I don't understand your point about extra processing time.  Canons shoot just as fast in mRAW and sRAW as in full RAW.

Semantic arguments aside about what does and does not constitute compression, using mRaw is just a different read off the sensor. It doesn't require any extra processing.  Raw data is maintained, albeit from fewer photosites.

Sorry, that's not true at all. RAW data is NOT maintained. These are YCC encoded formats (in this case, YCbCr). The sensor is read directly for luminance (Y), chrominance (Cb, Cr) reads DEMOSAIC THE PIXELS (!!),  and the information is processed to convert it into an entirely different format. The mRAW and sRAW formats utilize data from every pixel for the luminance channel (full luminance sampling). Color information for the Cr and Cb channels is interpolated from different pixel patterns (chrominance subsampling). The m/sRAW formats are 4:2:X pulldowns when it comes to encoding chrominance...4:2:2 in the case of mRAW and 4:2:0 in the case of sRAW.

This is very similar to how JPEG images are encoded, they also use a YCC variant (although one that preserves FAR less information). Its actually also very similar to how video frames are encoded.

There is no way, by any measure, that mRAW and sRAW can really be called "RAW" formats. They do not preserve the original sensor data in any way, shape, or form. The sensor is demosaiced for christ sake when reading and encoding chrominance channels.

For more information: http://dougkerr.net/Pumpkin/articles/sRaw.pdf

Osiris36

  • Guest
Re: More Big Megapixel Talk [CR1]
« Reply #93 on: September 26, 2012, 01:45:09 PM »
If the sensor is a 16-bit sensor with some kind of active cooling (no, not necessarily a fan), and Canon doesn't completely botch the ISO 100 and 200 electronic noise, then it could stomp all over the D800. With an extra two bits of information they could push 15 stops of DR, maybe even a little more (but no more than 16.0.)

My guess is that its still probably their same old sensor tech, but with some kind of efficient cooling to keep the sensor below room temperature (thereby reducing electronic noise), and extra bit depth. Canon needs the active cooling because they are either incapable of innovating and patenting technology similar but different enough to Sony Exmor, or there simply ISN'T another way to reduce noise electronically like Exmor, and Canon either has to pay Sony royalties, or do something entirely different.

It doesn't  matter if Canon choose 14 or 16 bit ADC as long the readout noise is so high as it is.
Canon with its old read out  technology and old sensor lines  can reduce the  readout noise slightly with slower read out, pictures/sec
Canon need Panasonic or Sony. (That Canon would buy sensors from Sony would be a huge loss of prestige)
Panasonic has the know how to use column-parallel analog-to-digital conversion
Panasonic has the column ADC technology Canon lacks.
Panasonic currently has a line of compact camera sensors - most of Canons compact cameras use Sony sensors.
Panasonic appears not to be able to make CMOS sensors that are as good as Canon at the pixel level.

So both companies can have benefit to work with each other

You don't really understand what you're saying if you're saying 16bit ADCs wouldn't make a difference.  16 bit ADCs in high speed mode will yield more reliable information than 14 bit ADCs if the input is anywhere near the 14 bit point from the pixel level data.  All that read noise everyone is thinking they are seeing is coming mostly from the ADCs being driven so fast with no headroom.  The column parallel ADC on sony/pana is good because:
1) they are very simple ADCs
2) they are very slow ADCs

Both of these are due to the pure numbers of them.  They are likely super conservative on both the count (probably way more than really needed) and performance (could be driven faster, but it's just cleaner this way).

Zlatko

  • Guest
Re: More Big Megapixel Talk [CR1]
« Reply #94 on: September 26, 2012, 01:54:42 PM »
m/sRAW is indeed much the same as a JPEG. It is a processed image format. Ironically, both use a YCC-base chromaticity/luminance storage structure, and both are compressed. The only real difference is that JPEG is lossy-compressed, where as m/sRAW are losslessly compressed. But none of them are actually "RAW"...they take the RAW data from the sensor, process it, and burn in the tone curves (picture style) and all the other camera settings into a final output image. You have more latitude in an m/sRAW image than a JPEG because it is not compressed with a lossy algorithm...so you have all the original YCC data, and because that original YCC data is 14bpc, rather than 8bpc.

You can push exposure around in post with m/sRAW a bit, and you can do moderate white balance corrections. But if you need the ability to do any significant editing in post (say, when you accidentally over- or under-expose a photo), your pretty much screwed if you use either one of those "RAW-but-not-really" formats. I've experimented extensively with them, and particularly with highlights and white balance, your freedom to correct is greatly limited relative to a true RAW.
Granted s/mRAW is not the same as full RAW, but your characterization of it as "much the same as JPEG" is not at all my experience.  I process 100's, usually 1,000's, of m/sRAW files every week and your characterization is not at all what I see.  If there is any less correct-ability than for full RAW, it is not meaningful in my work.  Perhaps your work is different.  I have all the latitude I need with m/sRAW files, but don't with JPEG.  With m/sRAW, I have made substantial adjustments to white balance and exposure many times with excellent results.  You can push them around in post more than just "a bit".

Zlatko

  • Guest
Re: More Big Megapixel Talk [CR1]
« Reply #95 on: September 26, 2012, 02:05:01 PM »
With all of the vulnerability you claim Canon is having, one would think you would be the first to be selling your Canon gear.  You say the D800 and D600 have superior sensors, and yet you haven't sold your Canon gear in order to buy Nikon?

these braindead arguments are very popular.

you canĀ“t criticise canon without some clown (with a rebel and a kit lense probably) jumping in the discussion telling you to sell your gear.   ::)
You can criticize Canon as much as  you wish, but when you make claims that Canon is failing and losing customers because of the supposed superiority of Nikon products, then you can expect to be asked why don't switch to the supposedly superior camera.  If it's so SUPERIOR, as claimed, then a switch would be the rationality and intelligent thing to do.  Who wouldn't want superior gear?  If a person makes such claims and states them as facts, but isn't switching, then that raises strong questions about their claims and their rationality.

bdunbar79

  • Canon EF 400mm f/2.8L IS II
  • *********
  • Posts: 3111
Re: More Big Megapixel Talk [CR1]
« Reply #96 on: September 26, 2012, 02:44:35 PM »
The reason why it isn't offered is because it takes extra processing to make a med raw file meaning way longer than a human can tak for a file size that large unless it had CFast 2 the new compact flash card system and stellar buffer.

Sorry, I don't understand your point about extra processing time.  Canons shoot just as fast in mRAW and sRAW as in full RAW.

You may want to Google this and review mRAW and sRAW creation.
2 x 1DX
B1G, MAC, GLIAC

canon rumors FORUM

Re: More Big Megapixel Talk [CR1]
« Reply #96 on: September 26, 2012, 02:44:35 PM »

jrista

  • Canon EF 600mm f/4L IS II
  • **********
  • Posts: 5334
  • EOL
    • Nature Photography
Re: More Big Megapixel Talk [CR1]
« Reply #97 on: September 26, 2012, 02:52:12 PM »
The reason why it isn't offered is because it takes extra processing to make a med raw file meaning way longer than a human can tak for a file size that large unless it had CFast 2 the new compact flash card system and stellar buffer.

Sorry, I don't understand your point about extra processing time.  Canons shoot just as fast in mRAW and sRAW as in full RAW.

You may want to Google this and review mRAW and sRAW creation.

I can't say I've experienced any marked degredation in write speed when using mRAW/sRAW. It isn't that much different than JPEG, same general processing however with more bits. JPEG can be written out at an ungodly rate, and that rate can be sustained nearly forever. You probably couldn't write out 150-200 m/sRAW continuous, but they shouldn't reduce your frame rate. The biggest drag on frame rate and buffer depth is the size of data WRITTEN to the memory card. The Digic chip is explicitly designed to handle the conversion processing, and shouldn't be imposing a noticeable drag.

V8Beast

  • 5DSR
  • *******
  • Posts: 1121
    • Stephen Kim Automotive Photography
Re: More Big Megapixel Talk [CR1]
« Reply #98 on: September 26, 2012, 02:57:39 PM »
How long until we hear about people who switched to Nikon wanting to switch back to Canon.....  :P

You won't because most of them never switched.

Zlatko

  • Guest
Re: More Big Megapixel Talk [CR1]
« Reply #99 on: September 26, 2012, 03:19:52 PM »
The reason why it isn't offered is because it takes extra processing to make a med raw file meaning way longer than a human can tak for a file size that large unless it had CFast 2 the new compact flash card system and stellar buffer.

Sorry, I don't understand your point about extra processing time.  Canons shoot just as fast in mRAW and sRAW as in full RAW.

You may want to Google this and review mRAW and sRAW creation.
Why Google it when my cameras create these files all of the time and there is no perceptible difference in write speed vs. full RAW?  The buffer clears in about the same time as when writing full RAW.  Is Google going to tell me that it takes 30-60 seconds to create a medium RAW file?  If so, it is wrong.

I'm pretty sure Nikon doesn't offer medium/small RAW because Canon has the patent, not because "it takes extra processing time".

HurtinMinorKey

  • Canon 6D
  • *****
  • Posts: 446
    • carolineculler.com
Re: More Big Megapixel Talk [CR1]
« Reply #100 on: September 26, 2012, 04:07:50 PM »
Sorry, that's not true at all. RAW data is NOT maintained. These are YCC encoded formats (in this case, YCbCr). The sensor is read directly for luminance (Y), chrominance (Cb, Cr) reads DEMOSAIC THE PIXELS (!!),  and the information is processed to convert it into an entirely different format. The mRAW and sRAW formats utilize data from every pixel for the luminance channel (full luminance sampling). Color information for the Cr and Cb channels is interpolated from different pixel patterns (chrominance subsampling). The m/sRAW formats are 4:2:X pulldowns when it comes to encoding chrominance...4:2:2 in the case of mRAW and 4:2:0 in the case of sRAW.

This is very similar to how JPEG images are encoded, they also use a YCC variant (although one that preserves FAR less information). Its actually also very similar to how video frames are encoded.

There is no way, by any measure, that mRAW and sRAW can really be called "RAW" formats. They do not preserve the original sensor data in any way, shape, or form. The sensor is demosaiced for christ sake when reading and encoding chrominance channels.

For more information: http://dougkerr.net/Pumpkin/articles/sRaw.pdf

There are no pixels on the sensor, only photo sites. The sRaw, and mRaw formats use information from fewer photo sites, but do not throw away any information from the photosites that are retained (or at least at the same level as regular Raw). The "pixels" are then generated from interpolation.

« Last Edit: September 26, 2012, 06:26:12 PM by HurtinMinorKey »

Blaze

  • Canon AE-1
  • ***
  • Posts: 71
Re: More Big Megapixel Talk [CR1]
« Reply #101 on: September 26, 2012, 06:09:39 PM »

There is no way, by any measure, that mRAW and sRAW can really be called "RAW" formats. They do not preserve the original sensor data in any way, shape, or form. The sensor is demosaiced for christ sake when reading and encoding chrominance channels.

For more information: http://dougkerr.net/Pumpkin/articles/sRaw.pdf


The article you linked to disagrees with you. The author claims that the sensor is not demosaiced, but rather that color four each pixel is computed from 2x2 RGGB blocks of photodetectors.

Relevant quote: " Thus, we can see that perhaps what happens to the Raw data on its way to the sRaw format does  not, rigorously, involve either demosaicing or downsampling."

jrista

  • Canon EF 600mm f/4L IS II
  • **********
  • Posts: 5334
  • EOL
    • Nature Photography
Re: More Big Megapixel Talk [CR1]
« Reply #102 on: September 26, 2012, 06:44:41 PM »

There is no way, by any measure, that mRAW and sRAW can really be called "RAW" formats. They do not preserve the original sensor data in any way, shape, or form. The sensor is demosaiced for christ sake when reading and encoding chrominance channels.

For more information: http://dougkerr.net/Pumpkin/articles/sRaw.pdf


The article you linked to disagrees with you. The author claims that the sensor is not demosaiced, but rather that color four each pixel is computed from 2x2 RGGB blocks of photodetectors.

ROFL. That IS demosaicing. Thats exactly what Adobe ACR, Lightroom, Aperture, etc. DO. They computer each RGB pixel from a 2xr RGGB blocks (MASAICS!) of photodecectors! In the case of software on the computer, the demosaicing is actually better, potentially a lot better, than the basic demosaicing done in-camera. ACR/LR use a pretty advanced form called Adaptive Homogeneity-directed Demosaicing, AHDD, to produce cleaner results than you get with m/sRAW. The article clearly describes demosaicing, although the author is being somewhat political by stating "it perhaps should not be called demosaicing." The only reason for that statement is the fact that the Y channel (luminance) is full resolution, and its only the Cr (red-green) and Cb (blue-green) channels that are actually demosaiced by processing multiple CFA pixels to produce the encoded output "pixel". Simple fact of the matter is, m/sRAW are not actually "raw, unmodified sensor data". They most definitely ARE modified, regardless of whether full luminance data is preserved or not.

Relevant quote: " Thus, we can see that perhaps what happens to the Raw data on its way to the sRaw format does  not, rigorously, involve either demosaicing or downsampling."

Again, note the use of "rigorously". Only the luminance channel is full resolution. Both chrominance channels are demosaiced and downsampled off the sensor. Depending on whether you are using s or m RAW, the amount of color data utilized for those channels differs...neither provide 100% color data like a true RAW image does.

You could call m/sRAW an intermediate format. It does not contain RGB pixels such that you could directly render it on a computer screen. It still needs further interpretation and interpolation. Its something between a RAW image and an RGB image. But it doesn't come close to a full RAW in terms of post-processing freedom and latitude...not by a long shot.

canon rumors FORUM

Re: More Big Megapixel Talk [CR1]
« Reply #102 on: September 26, 2012, 06:44:41 PM »

KitsVancouver

  • Rebel T6i
  • ****
  • Posts: 112
Re: More Big Megapixel Talk [CR1]
« Reply #103 on: September 26, 2012, 07:27:28 PM »
If they're prepared to drop the xD naming convention, you can be sure that it's price will top that of the current 1DX (which by the way is > $8k in Europe), so I wouldn't be surprised if it is $9,999 ($7,999 in USA market)

Agreed that if they are thinking of a new naming convention then the price is likely to be equal to or higher than the 1Dx. 

I'm starting to think more and more that there is no chance this thing is priced between the 5D Mark III and the 1Dx. 

HurtinMinorKey

  • Canon 6D
  • *****
  • Posts: 446
    • carolineculler.com
Re: More Big Megapixel Talk [CR1]
« Reply #104 on: September 26, 2012, 07:33:24 PM »
There is no specific luminance channel directly off the sensor. It's RGB photo sites, and the native ratio on all the canon sensors is 4-2-2. So big deal that mRaw is 4-2-2, so is Raw, no?

Each photosite gives us one 14bit measurement of one of the RGB channels. All this data is essentially the raw data file (with minamal processing).

mRaw and sRaw are just using fewer sites.  And just like full raw, they are demosaiced by DPP or whatever raw software you use.
« Last Edit: September 26, 2012, 07:37:42 PM by HurtinMinorKey »

canon rumors FORUM

Re: More Big Megapixel Talk [CR1]
« Reply #104 on: September 26, 2012, 07:33:24 PM »