Ad
Ad
Ad
Pages: « 1 [2] 3 4 ... 10 »   Bottom of Page
Print
Author Topic: This is why no RAW on the G7  (Read 69719 times)
bjanes
Sr. Member
****
Offline Offline

Posts: 2756



« Reply #20 on: December 03, 2006, 09:37:52 AM »
ReplyReply

Quote
I didn't make such a statement.  I said that ISO 1600 has less absolute noise, to distinguish it from relative noise, which is higher at ISO 1600 in all regards.
The photon shot noise is related only to the amount of light falling on the sensor; it has nothing to do with ISO, so when the same light falls on the sensor, ISO 100 and 1600 have the same shot noise, but the readout noise is greater at ISO 100, and hence the total of all absolute noises is higher for 100 (1600 is not worse in any case).
[a href=\"index.php?act=findpost&pid=88351\"][{POST_SNAPBACK}][/a]

Yes, but if you are using low ISO, you normally increase the exposure so that full well conditions exist for base ISO, collecting more photons and giving smaller SDs. At higher ISOs, the full well is not utilized and photon noise increases. If the light is very dim and normal exposure is not feasible, one can increase output in the RAW converter rather than increasing the ISO. According to Roger, there is no point of increasing the ISO substantially above unity gain, as we have discussed.

Quote
I will show you the results when I get to the experiment, but from what I have generally seen, bit depth is not as big a contributor to image DR as people make it out to be; it's the noise that makes the biggest difference in current systems.  Get noise out of the way, and then bit-depth will be the bigger issue.
[a href=\"index.php?act=findpost&pid=88351\"][{POST_SNAPBACK}][/a]

I would be interested in your experiments, since you know what you are doing. In another post, an electronics engineer who works with high end imaging also stated that few sensors can are really limited by 12 bit output.

Bill
Logged
BernardLanguillier
Sr. Member
****
Offline Offline

Posts: 7762



WWW
« Reply #21 on: December 03, 2006, 04:57:49 PM »
ReplyReply

Quote
That's because he's not doing any noise reduction whatsoever in his RAW workflow. I saw no mention of Neat Image, Noise Ninja, Noiseware, or any other PS noise reduction plugin. Obviously no noise reduction is done in-camera on RAW files, that's the whole point of RAW--giving the user control of what processing and how much, instead of the camera making all processing decisions. Thom's comparison is between in-camera noise reduction and nothing at all, and if the camera fails that test, the firmware programmers at Nikon ought to be fired en masse.
[a href=\"index.php?act=findpost&pid=88396\"][{POST_SNAPBACK}][/a]

Jonathan,

I think that he is comparing the jpg in camera, with the RAW produced by Nikon Capture, including noise reduction. That seems fair to me.

Most of the posters in this thread are saying that a PC processed version of a raw image will be better than the in camera jpg. I have assumed that people here were not extending the discussion to raw converters relative quality. From this standpoint, comparing the in camera jpg with the a raw file converted by the manufacturer's raw conversion software is the most reasonnable thing to do isn't it? Thom's result are very surprising, but they show that a jpg camera can be better noisewise than a raw conversion.

Not knowing exactly what the D50 built-in processor is able to do, I don't see how I could refute this fact based on general theoretical assumptions. For me his comparison is the best information we will ever get on this topic. I find it useful as one input in trying to understand Canon's move on the topic. Nothing more, nothing less.

As a final comment, Noise Ninja could be applied to both the jpg and raw.

Regards,
Bernard
Logged

A few images online here!
jani
Sr. Member
****
Offline Offline

Posts: 1604



WWW
« Reply #22 on: December 03, 2006, 05:08:47 PM »
ReplyReply

Quote
Not knowing exactly what the D50 built-in processor is able to do, I don't see how I could refute this fact based on general theoretical assumptions. For me his comparison is the best information we will ever get on this topic. I find it useful as one input in trying to understand Canon's move on the topic. Nothing more, nothing less.
It may be as simple as that Canon has no desire to have to provide similar noise reduction functionality in their raw converter (added complexity and support for the software division), but rather keep it in-camera.

And it is possible that the in-camera hardware and software might be better at specific noise handling than current Photoshop plugins or other noise handling software on a PC, even though that is likely to change fairly quickly as that software is updated.
Logged

Jan
John Sheehy
Sr. Member
****
Offline Offline

Posts: 838


« Reply #23 on: December 03, 2006, 05:35:20 PM »
ReplyReply

Quote
Yes, but if you are using low ISO, you normally increase the exposure so that full well conditions exist for base ISO, collecting more photons

Yes you do, but what does that have to do with the context of my original statement?


Quote
and giving smaller SDs.

The SDs are highest for full-well, and at ISO 100.  It is the ratio of the SD to the signal that is lowest.  If you take a RAW exposure of a color-checker card, and run a high-pass filter on it, the brightest square will have the most noise.

Quote
If the light is very dim and normal exposure is not feasible, one can increase output in the RAW converter rather than increasing the ISO.[a href=\"index.php?act=findpost&pid=88407\"][{POST_SNAPBACK}][/a]

You can do that, but what you have is really, in effect, a higher ISO.  The camera's ISO setting does not dictate the actual Exposure Index; absolute exposure and the brightness of the rendered image do.
Logged
Jonathan Wienke
Sr. Member
****
Offline Offline

Posts: 5759



WWW
« Reply #24 on: December 04, 2006, 08:58:03 AM »
ReplyReply

Quote
Jonathan,

I think that he is comparing the jpg in camera, with the RAW produced by Nikon Capture, including noise reduction. That seems fair to me.

Most of the posters in this thread are saying that a PC processed version of a raw image will be better than the in camera jpg. I have assumed that people here were not extending the discussion to raw converters relative quality. From this standpoint, comparing the in camera jpg with the a raw file converted by the manufacturer's raw conversion software is the most reasonnable thing to do isn't it? Thom's result are very surprising, but they show that a jpg camera can be better noisewise than a raw conversion.

Ignoring the numerous options available in RAW converters ignores one of the greatest advantages of RAW; being able to choose one's preferred tool based on the quality of the results produced by that tool, vs. being locked into the manufacturer's software, which in many cases is not that good, especially when it comes to noise reduction. Canon's RAW converter and ACR suck badly compared to Neat image in that department, which is why I convert in ACR with NR turned off and de-noise with Neat Image. Thom didn't specify what, if any noise reduction settings he was using in Nikon Capture, so assuming he did is an assumption, as is the premise that Nikon Capture's noise reduction is better than Neat Image, Noise Ninja, etc.

I want the freedom to use the best tool for the job as I see fit, not to be locked into one manufacturer's software that, in many cases, is crap. Neither Canon or Olympus' RAW converters are anything to write home about; they work, but ACR (especially when properly color calibrated) does a much better job with more accurate color.
Logged

BJL
Sr. Member
****
Offline Offline

Posts: 5121


« Reply #25 on: December 04, 2006, 10:00:02 AM »
ReplyReply

The Canon USA argument is a bit unpersuasive, because it seems to ignore some raw processing options like correcting white balance. Is it true that even when in-camera WB is applied, all relevant distinctions of luminance in each color channel are recorded in JPEG to the full accuracy limits of the sensor's noise levels, so that color adjustments in PP have as much significant information to work with from good JPEG as from raw data? Maybe.

Canon USA might well be correct a far as super-fine JPEG's 8-bit gamma compressed encoding of luminance data being able to record the full dynamic range and information content of the signal from those small photosites, so that there would be no advantage in that respect do doing raw conversion later on a computer.
Quote
Canon is claiming that the processor and firmware inside the camera can do a better job of processing the RAW data from the sensor than you, your much-more-powerful-than-the-camera desktop computer, your favorite RAW converter, and Photoshop.
[a href=\"index.php?act=findpost&pid=88311\"][{POST_SNAPBACK}][/a]
Not really: Canon USA is only claiming that you cannot do a better job by post-processing of raw than by post-processing of super-fine JPEG, because all the significant raw information (to the significance level imposed by noise) is preserved in that JPEG.

This is similar to another thread about whether there is any point doing A/D conversion at precision finer than one output level per electron of photo-site signal. The common question is whether the discretization (quantization) preserves all significant information or not. In this case, the discretization being to 8-bit gamma compressed JPEG values from what is probably 10 or 12 bit linear A/D convertor output.

My rough estimate is that the G7's sub-2 micron photo-sites have a well capacity of around 2,000 electrons or less and S/N ratio of about  1,000:1 or less, so 10-bit linear is the most that would be of value, except for those who like to measure noise to many bits of precision.

Does anyone know the bit-depth of the G7's A/D convertor?
Logged
jimhuber
Full Member
***
Offline Offline

Posts: 147


WWW
« Reply #26 on: December 04, 2006, 10:38:13 AM »
ReplyReply

Canon's position on the G7 is absolutely untenable. JPEG will never be better be than RAW. Even if it somehow magically were so at some point in time, RAW captures can be re-processed later with superior technology so the advantage can't possibly last.

Just this weekend my wife took my Canon S70 to a children's party and took some snapshots. One of them had terrible exposure due to a white table in the foreground getting blasted at close range by the flash. But with RawShooter Premium I processed it into two TIFF files that were identical except for exposure compensation, then blended them in PhotoShop - a ten minute rescue that would be impossible with JPEG.
Logged
bjanes
Sr. Member
****
Offline Offline

Posts: 2756



« Reply #27 on: December 04, 2006, 12:54:06 PM »
ReplyReply

Quote
Yes you do, but what does that have to do with the context of my original statement?
The SDs are highest for full-well, and at ISO 100.  It is the ratio of the SD to the signal that is lowest.  If you take a RAW exposure of a color-checker card, and run a high-pass filter on it, the brightest square will have the most noise.

[a href=\"index.php?act=findpost&pid=88492\"][{POST_SNAPBACK}][/a]

Yes John, I stand corrected on that point. The absolute noise is highest for ISO 100, but the S/N ratio is higher with the lower ISO. The context of the original statement is sometimes ambiguous: a discrepancancy could result from mispeaking, unclear communication, or confusion on my part. That is why followup is important.
« Last Edit: December 04, 2006, 04:49:21 PM by bjanes » Logged
BernardLanguillier
Sr. Member
****
Offline Offline

Posts: 7762



WWW
« Reply #28 on: December 04, 2006, 04:40:52 PM »
ReplyReply

Quote
Ignoring the numerous options available in RAW converters ignores one of the greatest advantages of RAW; being able to choose one's preferred tool based on the quality of the results produced by that tool, vs. being locked into the manufacturer's software, which in many cases is not that good, especially when it comes to noise reduction. Canon's RAW converter and ACR suck badly compared to Neat image in that department, which is why I convert in ACR with NR turned off and de-noise with Neat Image. Thom didn't specify what, if any noise reduction settings he was using in Nikon Capture, so assuming he did is an assumption, as is the premise that Nikon Capture's noise reduction is better than Neat Image, Noise Ninja, etc.
[a href=\"index.php?act=findpost&pid=88582\"][{POST_SNAPBACK}][/a]

Jonathan,

- Neat Image could be applied to the jpg image as well. The point is really what amount of noise/detail is present in the converted image vs jpg,
- Canon's decision if anything is probably based on their assessement of what their Raw converter can do. You might call this poor market intelligence if you'd like, but it makes complete sense from a Japanese standpoint. They just would have no control on the support of the G7 by third party software.

Quote
I want the freedom to use the best tool for the job as I see fit, not to be locked into one manufacturer's software that, in many cases, is crap. Neither Canon or Olympus' RAW converters are anything to write home about; they work, but ACR (especially when properly color calibrated) does a much better job with more accurate color.
[a href=\"index.php?act=findpost&pid=88582\"][{POST_SNAPBACK}][/a]

I understand Jonathan, but to what extend has this influenced Canon's decision?

Cheers,
Bernard
Logged

A few images online here!
bjanes
Sr. Member
****
Offline Offline

Posts: 2756



« Reply #29 on: December 04, 2006, 04:58:04 PM »
ReplyReply

Quote
Ignoring the numerous options available in RAW converters ignores one of the greatest advantages of RAW; being able to choose one's preferred tool based on the quality of the results produced by that tool, vs. being locked into the manufacturer's software, which in many cases is not that good, especially when it comes to noise reduction. Canon's RAW converter and ACR suck badly compared to Neat image in that department, which is why I convert in ACR with NR turned off and de-noise with Neat Image. Thom didn't specify what, if any noise reduction settings he was using in Nikon Capture, so assuming he did is an assumption, as is the premise that Nikon Capture's noise reduction is better than Neat Image, Noise Ninja, etc.

[a href=\"index.php?act=findpost&pid=88582\"][{POST_SNAPBACK}][/a]

With JPEGs it is not possible to turn off NR completely with the D200 and some other Nikon cameras when using high ISOs. Nikon Capture and ACR do allow NR to be disabled. I don't think that anyone claimed that NC or ACR or in camera has better NR than Neat Image or NN. If you want to use separate NR, it is important to turn off all sharpening for the JPEG and as much NR as possible.

At high ISO, I think you would get better results with raw for conversion and NN or NI for NR. Personally I hardly ever use JPEGs, but many are satisfied with its results and the differences are minimal in many cases.
« Last Edit: December 04, 2006, 05:14:46 PM by bjanes » Logged
Jonathan Wienke
Sr. Member
****
Offline Offline

Posts: 5759



WWW
« Reply #30 on: December 04, 2006, 05:27:51 PM »
ReplyReply

Quote
Jonathan,

- Neat Image could be applied to the jpg image as well. The point is really what amount of noise/detail is present in the converted image vs jpg,
- Canon's decision if anything is probably based on their assessement of what their Raw converter can do. You might call this poor market intelligence if you'd like, but it makes complete sense from a Japanese standpoint. They just would have no control on the support of the G7 by third party software.
I understand Jonathan, but to what extend has this influenced Canon's decision?

Cheers,
Bernard

I don't believe Canon's assertions regarding RAW vs superfine JPEG noise are correct; they fly in the face of all of my experience processing RAW and JPEG images and the relative quality of the two. Yes, noise reduction can be applied to JPEGs, but doing so is less effective than applying the same noise reduction tool on a 16-bit converted RAW. There are no JPEG artifacts in the RAW data, and noise reduction tools work better in 16-bit mode, as do lens correction, sharpening, and other tools.

Nor do Canon's assertions address white balance or color accuracy issues at all; a properly profiled/calibrated RAW converter is clearly superior to camera JPEGs in both color accuracy and ease of setting the correct white balance, especially when the camera doesn't get WB quite right. Canned printer profiles are not better than well-made custom profiles in the vast majority of cases, why should the generic camera profiles used to create camera JPEGs be better than a custom-profiled/calibrated RAW converter? The notion simply defies common sense.

And what about tone curves? With all decent RAW converters, you have the ability to select or create the most appriate tone curve for the image after the shot, and preview the results of tweaks as you do so, to ensure the best possible result. With camera JPEGs, if you don't like the camera-applied tone curve, you are far more limited in how much you can change it before banding and posterization set in.

I don't really care what Canon is basing their decisions on, or whether those decisions make sense to the Japanese mindset; the bottom line is that the decisions are stupid, based on premises of extremely dubious merit, and very much likely to alienate the serious-photographer-wanting-a-decent-compact-camera market segment. They've certainly alienated me; I bought the Olympus SP-350 specifically because it supports RAW. I would have preferred to buy a Canon instead to maintain E-TTL flash compatibility and not have to buy yet another memory card format, but Canon didn't see fit to offer me what I was looking for, so I went elsewhere.

Even if camera JPEGs were always just as good as the output of the best RAW converter + noise reduction tools in existence (which they are certainly not, for the reasons I mention above), adding the ability to save a RAW file to firmware functionality is trivial; certainly far less complex than converting the RAW data to JPEG and then saving it. There is no logically defensible argument for not allowing the RAW data to be saved.

I realize I'm just one person. But if enough of us complain loudly enough, the current silliness may go away.
Logged

BernardLanguillier
Sr. Member
****
Offline Offline

Posts: 7762



WWW
« Reply #31 on: December 04, 2006, 06:11:56 PM »
ReplyReply

Quote
I realize I'm just one person. But if enough of us complain loudly enough, the current silliness may go away.
[a href=\"index.php?act=findpost&pid=88670\"][{POST_SNAPBACK}][/a]

Jonathan,

I am completely with you in wishing that Canon had kept RAW support in the G7, and most people here at LL probably feel the same, but from a forum standpoint that's the scope of another thread IMHO.

Regards,
Bernard
Logged

A few images online here!
howiesmith
Guest
« Reply #32 on: December 04, 2006, 06:49:50 PM »
ReplyReply

In my opinion, some folks on this sight have lost sight that the G7 is a point and shoot camera, not a professional tool.  For every person who screams the G7 should (or must) have RAW, I cam point to 10 camera buyers who say "What's RAW?  White balance?"

All the consumer (who is the alledged audience for this camera) seem to want is a really nice camera that takes "clear" pictures, is easy to use and makes a decent print or e-mail attachment of the family dog (or kid).

Reading this thread, I do not get the idea that Canon said JPEGs were better than RAW, there just wasn't enough difference.

I also think that if everyone complaining on LL made a loud enough cry and actually bought a G7 with RAW, Canon would never notice.  But they might notice the loss of sales because the "pricey" G7 (per dpreviw) were even pricier and a lot of consumers went elsewhere.

Canon may have screwed up, but maybe they should be given credit for knowing their business.
Logged
BernardLanguillier
Sr. Member
****
Offline Offline

Posts: 7762



WWW
« Reply #33 on: December 04, 2006, 07:07:34 PM »
ReplyReply

Quote
In my opinion, some folks on this sight have lost sight that the G7 is a point and shoot camera, not a professional tool.  For every person who screams the G7 should (or must) have RAW, I cam point to 10 camera buyers who say "What's RAW?  White balance?"
[a href=\"index.php?act=findpost&pid=88687\"][{POST_SNAPBACK}][/a]

You are of course right, but many of us had hoped that Canon would keep targetting advanced users with their G series. Let's face it, we all have back problems, and carrying a D80 in the bag at all times isn't realistic.

We are now facing the hard reality that we are not a large enough population, and that probably no-one will ever manufacture the camera we had hoped for... that's a bit like learning that Santa Claus isn't real, life is tough...

Cheers,
Bernard
Logged

A few images online here!
howiesmith
Guest
« Reply #34 on: December 04, 2006, 09:02:02 PM »
ReplyReply

Quote
We are now facing the hard reality that we are not a large enough population, and that probably no-one will ever manufacture the camera we had hoped for... that's a bit like learning that Santa Claus isn't real, life is tough...

Cheers,
Bernard
[a href=\"index.php?act=findpost&pid=88692\"][{POST_SNAPBACK}][/a]

Only thing tougher than not being the big fish in the pond is realizing you are really just a very little fish in the little pond.
Logged
Jonathan Wienke
Sr. Member
****
Offline Offline

Posts: 5759



WWW
« Reply #35 on: December 05, 2006, 01:23:10 AM »
ReplyReply

Quote
also think that if everyone complaining on LL made a loud enough cry and actually bought a G7 with RAW, Canon would never notice.  But they might notice the loss of sales because the "pricey" G7 (per dpreviw) were even pricier and a lot of consumers went elsewhere.

Canon may have screwed up, but maybe they should be given credit for knowing their business.

Adding RAW support to the firmware functionality is a trivial programming exercise that at most would make a few cents' difference in the cost of the camera over its life cycle.
Logged

BernardLanguillier
Sr. Member
****
Offline Offline

Posts: 7762



WWW
« Reply #36 on: December 05, 2006, 03:56:44 AM »
ReplyReply

Quote
Adding RAW support to the firmware functionality is a trivial programming exercise that at most would make a few cents' difference in the cost of the camera over its life cycle.
[a href=\"index.php?act=findpost&pid=88729\"][{POST_SNAPBACK}][/a]

Isn't this coming from the same company that purposedely disabled some firmware functions in the 300D to create a differentiation with its bigger brothers?

i remember some Russian computer genious hacking this and providing back the functionalities...

Cheers,
Bernard
Logged

A few images online here!
eronald
Sr. Member
****
Offline Offline

Posts: 3870



WWW
« Reply #37 on: December 05, 2006, 05:51:37 AM »
ReplyReply

Am I the only one who wants a pro-quality point and shoot ? SLR-quality sensor, carbon-fiber body, Raw, a decent wide-angle zoom, and all the usual automation.  And small. Leica, what are you waiting for ?

Edmund
Logged

Edmund Ronald, Ph.D. 
opgr
Sr. Member
****
Offline Offline

Posts: 1125


WWW
« Reply #38 on: December 05, 2006, 06:42:00 AM »
ReplyReply

Quote
Adding RAW support to the firmware functionality is a trivial programming exercise that at most would make a few cents' difference in the cost of the camera over its life cycle.
[a href=\"index.php?act=findpost&pid=88729\"][{POST_SNAPBACK}][/a]

I wouldn't be surprised though if the RAW file doesn't compress that well, and writing it to the memory card will take long, and will fill it up real quick. I wonder if they could do it in less than 6sec, and I also wonder what people consider useable raw:

1raw in 6sec obviously doesn't cut it,
3raws in 3secs?

It's easy to write these seemingly small numbers, but I strongly suspect that the impact on design requirements is more than significant, even at the end of 2006... :-(
Logged

Regards,
Oscar Rysdyk
theimagingfactory
Jonathan Wienke
Sr. Member
****
Offline Offline

Posts: 5759



WWW
« Reply #39 on: December 05, 2006, 08:07:32 AM »
ReplyReply

Quote
I wouldn't be surprised though if the RAW file doesn't compress that well, and writing it to the memory card will take long, and will fill it up real quick.

Canon CR2 RAW files losslessly compress to about 1/2 to 2/3 of uncompressed size, depending on ISO and subject. DNG does just a bit better than that.
Logged

Pages: « 1 [2] 3 4 ... 10 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad