Photo Forum / Digital Photography / DSLR Cameras / September 2006
What's the big deal with ISO 3200 Boost?
|
|
Thread rating:  |
plastic_razor@yahoo.com - 04 Sep 2006 11:48 GMT Whenever dSLRs get into a measuring contest with each other, the ISO 3200 "boost" is always being touted as a feature by cameras that have it. For example, I've read on several product reviews that one of Canon 30D's advantages over the puny Digital Rebel is this so-called ISO 3200 Boost. Even the most reputable reviewers are including ISO 3200 in their graph comparisons.
Am I wrong in thinking that this "boost" is nothing more than software manipulation? Similar to how "Point n' Shoot" cameras are constantly peddling their "digital zooms" to anybody that would listen.
Correct me if I'm mistaken, but in the case of Digital Rebel XT vs 30D, their performance over the entire spectrum (ISO 100 to 1600) are virtually identical. And if this "ISO 3200 Boost" really is fool's gold (?), that would make these two cameras equal performers in this particular aspect of digital photography... correct?
w.beckley@gmail.com - 04 Sep 2006 12:11 GMT > Whenever dSLRs get into a measuring contest with each other, the ISO > 3200 "boost" is always being touted as a feature by cameras that have [quoted text clipped - 12 lines] > gold (?), that would make these two cameras equal performers in this > particular aspect of digital photography... correct? My take is this: I've used the boost on my 20D to great success, and gotten great (usable) shots using it that I'd not have had if I stayed at 1600.
That said, this boost is a small reason to pick the 20D and 30D over the XT (and now XTi). The build quality and ergonomics far outweigh the gain in ISO (or resolution in the case of the XTi). The 20D and 30D are solid pieces of kit. If small size isn't the most important thing you've got, you'd do well to grab the 30D. Its ergonomics alone make it worthwhile. The build quality seals it. 3200 ISO is just a bonus (but it does work and is great to have- hardly "fools gold").
That's me, but I love low light stuff...
Will
RichA - 04 Sep 2006 17:50 GMT > Whenever dSLRs get into a measuring contest with each other, the ISO > 3200 "boost" is always being touted as a feature by cameras that have [quoted text clipped - 12 lines] > gold (?), that would make these two cameras equal performers in this > particular aspect of digital photography... correct? If you take an image shot at 800 ISO and one shot at 3200, and equalize the lighting, you'll find that the 800 ISO image can retain as much detail with less noise but the 3200 preserves contrast and colour better in darkened areas. However, I'm not sure that 3200 boosting is the same as having a "real" 3200 ISO, as the image illumination (compared to 1600 ISO) does not seem to suggest a full stop increase. Though technically, everything beyond 100 or 200 ISO is "boosted" from what I've read. Cameras with sensors with large pixels probably do better here.
Marc Sabatella - 04 Sep 2006 18:19 GMT > Am I wrong in thinking that this "boost" is nothing more than software > manipulation? Similar to how "Point n' Shoot" cameras are constantly > peddling their "digital zooms" to anybody that would listen. That's a bit harsh, I think. It is true that you can shoot at a lower ISO and then push it in PP, and get results that will probably be as good as the 3200 that may be provided on camera - at least, if you shoot RAW. But it can be useful to being able to see the image at the "correct" exposure on the LCD - it's much easier to judge whether the shot came off well enough in terms of sharpness etc. That said, I seldom use ISO 3200, because it's still way too noisy on my Pentax *istDS (which doesn't *call* it a "boost", but as someone else says, in some sense, anything above the native ISO of the sensor is just a boost of one sort or another).
--------------- Marc Sabatella marc@outsideshore.com
Music, art, & educational materials Featuring "A Jazz Improvisation Primer" http://www.outsideshore.com/
default - 04 Sep 2006 20:29 GMT > Correct me if I'm mistaken, but in the case of Digital Rebel XT vs 30D, > their performance over the entire spectrum (ISO 100 to 1600) are > virtually identical. And if this "ISO 3200 Boost" really is fool's > gold (?), that would make these two cameras equal performers in this > particular aspect of digital photography... correct? I recall reading somewhere that if you look into a ISO3200 RAW file from a Canon camera that you will find only even numbers suggesting the values are simply doubled from a ISO 1600 exposure, 1 stop underexposed. If this is truly the case, then you can get pretty much identical results from your Rebel XT by sliding the exposure compensation down and stop and pushing by one stop in your RAW converter.
This works well for me with my XT when I am already at ISO1600 and desperately want more shutter speed. It's especially helpful when you are in low light with a telephoto hand-held and weren't prepared. You get the added benefit of more headroom for highlights too that you can compress in. If you use the noise reduction in the new Canon Digital Photo Professional or Adobe Camera RAW, you can get a quite nice picture as long as you don't expect to blow it up huge and way better than high-speed 35mm film in the same circumstance.
I don't think ISO3200 is a bunch of pyrite but just a convenience that Canon through in for the 10/20/30D cameras.
Paul Furman - 04 Sep 2006 23:50 GMT > Whenever dSLRs get into a measuring contest with each other, the ISO > 3200 "boost" is always being touted as a feature by cameras that have [quoted text clipped - 12 lines] > gold (?), that would make these two cameras equal performers in this > particular aspect of digital photography... correct? I'd guess there's a slight improvement over boosting RAW files, easy enough to test, mostly it's just more convenient when you need shutter speed for low light, motion or DOF for macros.
 Signature Paul Furman http://www.edgehill.net/1 Bay Natives http://www.baynatives.com
Roger N. Clark (change username to rnclark) - 05 Sep 2006 04:16 GMT > Whenever dSLRs get into a measuring contest with each other, the ISO > 3200 "boost" is always being touted as a feature by cameras that have [quoted text clipped - 6 lines] > manipulation? Similar to how "Point n' Shoot" cameras are constantly > peddling their "digital zooms" to anybody that would listen. In early DSLRs it was reportedly a mathematical boost, but it newer cameras, it is not.
> Correct me if I'm mistaken, but in the case of Digital Rebel XT vs 30D, > their performance over the entire spectrum (ISO 100 to 1600) are > virtually identical. And if this "ISO 3200 Boost" really is fool's > gold (?), that would make these two cameras equal performers in this > particular aspect of digital photography... correct? There is a simple upper limit beyond which you won't gain anything. Each camera's electronics had a gain stage before feeding the A/D, and the gain is expressed in electrons/DN where DN is "data number," There is no point in having gain above 1 electron. Most DSLRs have that gain point between the iso 800 and iso 1600 levels. For example, see Table 1, column E at:
Procedures for Evaluating Digital Camera Sensor Noise, Dynamic Range, and Full Well Capacities; Canon 1D Mark II Analysis http://www.clarkvision.com/imagedetail/evaluation-1d2
Other sensors can be found at: http://www.clarkvision.com/imagedetail and scrolling down to sensor analysis. I'll be adding several other cameras in the next few weeks. For example, preliminary analysis of the Canon S70 7-megapixel camera shows its gain at minimum is about 2 electrons/DN at ISO 50, so the gain =1 at iso 100.
Minimum noise in cameras is around 4 electrons (called the read noise), so at a gain =1, read noise is well digitized. There is no need to go higher in a camera, as you can get the same result in post processing and have a higher dynamic range.
I've tabulated gain data of mine and from the net here: http://www.clarkvision.com/imagedetail/digital.signal.to.noise in Table 1. The 30D has the same sensor as the 20D, and probably the same gains. The 20D gain is 1.5 at ISO 800, and 0.8 at iso 1600. There is no advantage going above iso 1600 unless you want to not post-process images.
Amateur astronomers using these cameras have independently found that iso 3200 has no advantage. Most use ISO 1600 or 800.
Roger
Paul Furman - 06 Sep 2006 16:18 GMT <snip>
> Amateur astronomers using these cameras have independently > found that iso 3200 has no advantage. Most use ISO > 1600 or 800. OK thanks Roger, so it's been tested and is simply a convenience for jpeg shooting.
 Signature Paul Furman http://www.edgehill.net/1 Bay Natives http://www.baynatives.com
Protoncek (ex.SleeperMan) - 06 Sep 2006 20:15 GMT >> Whenever dSLRs get into a measuring contest with each other, the ISO >> 3200 "boost" is always being touted as a feature by cameras that have [quoted text clipped - 55 lines] > > Roger could you explain why there would be no advantage in using 3200? If you shoot at 1600, you have to either shoot with longer speed or with lower F number. longer shutter speed will give you blurry pic, which hundred PS programs won't remove, while you (most of the time) already shoot with lowest F possible, so nothing you can do here. SO your only option would be underexposed image, while lightening in PS will indoubtly result in even more noise, so i'd say wou'll still get less noise at 3200 then at 1600 and compensate exposure setting in PS.
Paul Furman - 07 Sep 2006 16:47 GMT > "Roger N. Clark (change username to rnclark)" <username@qwest.net> wrote > >>Amateur astronomers using these cameras have independently >>found that iso 3200 has no advantage. Most use ISO >>1600 or 800. > could you explain why there would be no advantage in using 3200? If you
> shoot at 1600, you have to either shoot with longer speed or with lower F > number. longer shutter speed will give you blurry pic, which hundred PS [quoted text clipped - 3 lines] > more noise, so i'd say wou'll still get less noise at 3200 then at 1600 and > compensate exposure setting in PS. Amateur astronomers do use something a bit different than a standard RAW converter and it's a different game, so there may be some advantage for 'normal' photography. It would be a pretty simple test to compare underexposed ISO 1600 boosted in RAW to ISO 3200 at normal exposure.
 Signature Paul Furman http://www.edgehill.net/1 Bay Natives http://www.baynatives.com
Protoncek (ex.SleeperMan) - 07 Sep 2006 20:28 GMT >>>Amateur astronomers using these cameras have independently >>>found that iso 3200 has no advantage. Most use ISO [quoted text clipped - 13 lines] > 'normal' photography. It would be a pretty simple test to compare > underexposed ISO 1600 boosted in RAW to ISO 3200 at normal exposure. that should be easy to do... it's just these days i did a lot of experimenting and what i noticed at all times is that when working with underexposed photo, noise really increases quite a lot. I'll do that (tomorrow, i guess) and see...
w.beckley@gmail.com - 08 Sep 2006 10:09 GMT > could you explain why there would be no advantage in using 3200? If you > shoot at 1600, you have to either shoot with longer speed or with lower F [quoted text clipped - 4 lines] > more noise, so i'd say wou'll still get less noise at 3200 then at 1600 and > compensate exposure setting in PS. I think that the theory is that is ISO "boost" to 3200 is as rough in-camera as some suspect, then the advantage is that Adobe's Camera RAW (or Lightroom) or PhaseOne's CaptureOne software does a better job of extracting more detail than the in-camera solution from Canon. Which is certainly reasonable. Think about it... if you underexpose at 1600 by one stop you're exposing the same way you would at 3200. The question is, do you trust the DiGiC2 processor and Canon's in-camera software more than you trust Adobe or PhaseOne to bring that exposure into a usable place? More to the point, as others have said, one could easily test this and find out which one is resulting in better images. Both options seems real to me. I've gotten great results from the ISO boost in the 20d, but at the same time, I haven't done extensive testing against 1600.
Will
Protoncek (ex.SleeperMan) - 08 Sep 2006 14:25 GMT >> could you explain why there would be no advantage in using 3200? If you >> shoot at 1600, you have to either shoot with longer speed or with lower F [quoted text clipped - 22 lines] > > Will hm...i doubt that underexposing at 1600 equals correct exposing at 3200...since when you increase ISO, you basically increase sensor sensitivity, so sensor will give same data at lower light and worse settings---like, say, F4, 1/60, ISO1600 and F4, 1/125, ISO3200 should be around the same, with somewhat increased noise at 3200,sure. I mean .- digitally same...since at one certain point would you measure, say, number 125 (on range 0-255 digitally) on 1600 or 3200, while ehen underexposing you would get number, say, only 35. Increasing that number to 125 (= correcting underexposure) would also increase a lot of noise.
From my point this is more or less the same as it is in audio amplifier status: it's always better to have more amplifier stages with lower gain then one with high, since when you increase gain of one transistor, you add noise, also. And since noise curve is not linear, we can assume one number of noise at certain gain, and NOT double noise at double gain, but at least triple noise at double gain.
That's my point of view....correct me if i'm wrong. But i will try to do some test shots this evening and if all goes ok, i'll put then on my page and post links...just of curiosity..
DP - 09 Sep 2006 06:04 GMT > From my point this is more or less the same as it is in audio amplifier > status: it's always better to have more amplifier stages with lower gain > then one with high, since when you increase gain of one transistor, you > add noise, also. Well, my audio amp's volume control goes up to 11, so I'm cool.
Doug McDonald - 08 Sep 2006 15:40 GMT > I think that the theory is that is ISO "boost" to 3200 is as rough > in-camera as some suspect, then the advantage is that Adobe's Camera [quoted text clipped - 9 lines] > boost in the 20d, but at the same time, I haven't done extensive > testing against 1600. I have a new 30D and tried comparing 1600 and 3200 RAW files made at the same manual settings, comparing the raw files in Photoshop. My tentative conclusion is, no noticeable difference. However, on the on-camera review screen, the 1600 ones looked underexposed. I think it is a tempest in a teapot.
Doug McDonald
Protoncek (ex.SleeperMan) - 08 Sep 2006 18:30 GMT >> I think that the theory is that is ISO "boost" to 3200 is as rough >> in-camera as some suspect, then the advantage is that Adobe's Camera [quoted text clipped - 17 lines] > > Doug McDonald i tried similar ---set manual exposure and shot at 1600 and 3200. Settings were ideal for 3200, so surely image was underexposed at 1600. But, the catch is that when you open an image in ACR, auto settings are chosen so that image comes out pretty good, in my case the one shot at 1600 was automatically corrected +1,2 exposure, so end images were very similar. Also i didn't find any noticeable difference in noise, so i start to wonder where is the catch...it surely must be, since i bet that 30d is so different than 350d that Canon wouldn't go for this "small " catch.. or would it?
|
|
|