350D/XT dynamic range

[randalcandari]

hi guys, just a quick question, about how many stops is your 350D's dynamic range? and what is the safe range to avoid overexposing/underezposing some parts of the image? thanks!

-randal

[scottbergerphoto]

I believe the dr of most of Canon's new cmos sensors is 8-10 stops.

[Nightcrawler]

I thought I remember seeing a website that rated the 10D at 10-11 stops. I don't know where the XT would be though. I imagine it wouldn't be much different.

[etaf]

these may help
http://www.photo.net/bboard/q-and-a-fetch-msg?msg_id=00CMSr
http://clarkvision.com/imagedetail/dynamicrange2/
http://www.canon.com.sg/index.cfm?fuseaction=digitalcamera&prod_type=technology
http://www.photo.net/learn/drange/

[randalcandari]

If the dynamic range is 8-10 steps, would that mean that if a meter reading gives the following at f5.6, 1/4, 1/8, 1/25, 1/30, and 1/80, and I set the shutter speed at 1/10sec then everything should be just be exposed properly right? or I'm misinterpreting everything. I noticed that when I do the settings above, the 1/80 areas are blown out and the 1/4 are very dark. Is this common to DSLRs or is my unit defective? by the way I'm using an XT.

Thanks!

[tzalman]

randalcandari -
DR which is technically (mathematically) present is a different thing from that which is perceptually present. If we assume for a moment that the digital sensor can capture a 10 stop range then we can compare it to the classic Zone System - although only approximately because Adams was talking about film and silver coated paper. The Zone System describes ten zones. from 0 (absolute black) to IX (absolute white). Zone V is the medium grey that all light metering systems strive to acheive. However, because of the nature of human vision and the limitions of our hardware (silver paper, inkjet printers, monitors) between Zones 0, I and II differences can barely be seen. Zone III is called "The Darkest Detailed Shadow." The same scrunching together is true at the top of the scale, so Zone VII is described as "Textured Highlight" and Zone VIII is so close to white that it appears blown. So instead of ten stops practical latitude we have only seven. Note that the difference between Zone V and Zone VII is two stops. Thus if the brightest highlight in which you want to retain detail spot-meters at 1/80 your exposure should be no greater than 1/20 (and not 1/10 as you wrote). At the shadow end your RAW file may contain data that will have to be edited in order to make it visible.

[ron chappel]

Randalcandari
Sorry to tell you but there is nothing wrong with your camera-it's completely normal for canon DSLR's to give a 5-6 stop dynamic range in normal use.
I see this wrongly quoted SO often and it does get abit annoying (not to mention the effect it has on poor owners who think they are doing something wrong)
Anybody who is getting more than 6 stops this is doing one of the following:

Double exposure methods such as shooting RAW then making two frames (one bright,one dark) the recombining them in photoshop.This method gives a usefull one stop or more increase but does give noisier shaddows depending how far it's taken.
I've seen examples of this on the net where it's taken to rediculous extremes-claims of huge DR ..but they are counting individual bright pixels in the shaddows that are otherwise completely dark.No use in real life whatsoever

True multiple exposure methods-where several differently exposed images are combined.This gives perfect images with an unlimited DR increase...but of course can only be used for non moving subjects

[Movick]

Agreed,

Digital SLRs have approx 5-6 stops of dynamic or equivalent film contrast range (essentially the same that of color negative film. B&W film has the greatest contrast range of all current mediums at 7 stops. The aforementioned 10-11 stop DSLR range is a misnomer. The only way to achieve a full tonal range of 11 stops (pure black to pure white is to expose in 2 separate images for highlights and lowlights and combine the layers.

Movick

[Pekka]

Acoording to http://www.normankoren.com/digital_tonality.html the total dynamic range of the EOS-10D is 8.5 f-stops. Chuck Westfall mentioned that 1D Mark II has about 9 stops.

[randalcandari]

Thanks everyone for your replies. things like this makes this hobby more interesting.
So if the meter gives 1/10sec for the brightest area and I try to shoot at 1/2sec, its normal for the bright areas to be blown out?

[ron chappel]

Ok,where these shutter speed numbers cross over from fractions of a sec to seconds+decimals on my camera completely confuses me early in the morning
....but i think 1/10th to 1/2 is two and a third stops

If that is right it probably shouldn't be blowing out (just).Are the metering measurements acurate-are you using spot metering?

[Movick]

<<Acoording to http://www.normankoren.com/digital_tonality.html the total dynamic range of the EOS-10D is 8.5 f-stops. Chuck Westfall mentioned that 1D Mark II has about 9 stops.>>

There's no way that a DSLR's tonal range can exceed that of B&W film, which as I've already mentioned is only 7 stops. Nine stops will veritably exceed the entire contrast range of pure black to pure white. Take a shot of a correctly lit pure black card or paper and a pure white sheet of same set side by side, with your MKII and see where your tonal values fall. One of the yin/yang elements will have to be compromised tonally to correctly expose the other.



I slapped together the attached 7 stop tonal ruler to exemplify this point.


Movick

[randalcandari]

Ok,where these shutter speed numbers cross over from fractions of a sec to seconds+decimals on my camera completely confuses me early in the morning
....but i think 1/10th to 1/2 is two and a third stops

If that is right it probably shouldn't be blowing out (just).Are the metering measurements acurate-are you using spot metering?

yes those are fractions, i convert decimals to fractions to avoid confusions. I'm using partial metering because spot metering is not available in 350D/XT. is there a huge difference between those two metering modes?

Thanks!

[Pekka]

<<Acoording to http://www.normankoren.com/digital_tonality.html the total dynamic range of the EOS-10D is 8.5 f-stops. Chuck Westfall mentioned that 1D Mark II has about 9 stops.>>

[color=black]There's no way that a DSLR's tonal range can exceed that of B&W film

Why? Film has more noise and digital technology has improved a lot lately. Paper is out of digital equation.

Nine stops will veritably exceed the entire contrast range of pure black to pure white. Take a shot of a correctly lit pure black card or paper and a pure white sheet of same set side by side, with your MKII and see where your tonal values fall. One of the yin/yang elements will have to be compromised tonally to correctly expose the other.

Film? Paper?

Digital cameras CAPTURE a scene in bits (sensor data array). Mark II sensor captures 12 bits per channel which can store about 9 stops. The DIGIC II chip processes that data and stores that information into RAW file. After that you can use your preferred application available to convert it to e.g. 48 bit TIFF, using curves and tools to expand the data to your working gamma and color space (e.g. Gamma 2.2 ProPhoto RGB). So 8.5 stops DR in final image (.5 goes to noise) is not impossible.

So, when you say in film-tongue if you expose this target for full white the black is not black and vice versa, and that proves dynamic range of device Y is not more than X stops, in digital-tonque you capture the information in target and later use tools to EXPAND or COMPRESS the data and save it in format suitable for the TARGET media (of course in bounds of artistic requirements). The more bits you have per channel the smoother the expansion/compression result will be. RAW software can interpolate, too, and I suspect most of them do.

[robertwgross]

From that diagram, what does 144% reflective mean? That math worries me.

---Bob Gross---

[Tom W]

There seems to be a number of ways in which dynamic range can be measured, and that is a problem. The 7-stop chart may be useful from a printing point of view, but how much brighter is a light bulb compared to pure black in terms of stops? How about the sun? Photography hasn't the ablility to light beyond pure white, and printing can only go as white as the paper allows.

Of course, the dynamic range of the camera can go beyond those extremes if it can allow recover of detail from shadows beyond those which are seen on the screen, or pull down blown (or almost blown) highlights to reveal details in the high-to-over exposed portions of the image. And that's what the dynamic range is about - the ability to move the exposure around a bit after-the-fact to show that parts of the image that may be hidden at the extremes. Remember that all that end-to-end detail still needs to be squeezed into the narrower dynamic range in which the monitor or printer/paper can display.
This can be a fascinating subject.

[slin100]

Tom makes several good points. The contrast range of a scene can easily exceed 10000:1 (more than 13 stops) and go up to 100000:1 or more. OTOH, that much contrast is pretty taxing to the human visual system, which, I read somewhere, can only comfortably process at most 10000:1 contrast range. That's different from the sensitivity of the human visual system, which perceive everything from starlight to the sunlight, which is easily a difference of 100000000:1, just not all at once.

[ron chappel]

Movick you have made several massively wrong assumtions and plain obvious mistakes in your argument.I won't go into detail,it's not worth the effort.The previous posters have mentioned the main mistakes.
I'd advise some serious study into this subject.Starting with the EV scale would be a good first step

[randalcandari]

hey guys before we go on super technical details, please help me. The main reason I post this is because I'm suspecting my XT has a problem in its dynamic range, I think it's to narrow. normally I use partial metering since that's the only mode available on XT closest to spot metering (I think). I notice that bright areas on a scene gets washed out easily even if the camera measures only about 2 2/3 stops of dynamic range. for example, in Av mode partial metering, i get a 1sec for the dark areas and 1/8sec on bright areas. that's just about 3 stops right, when I set the camera to expose the scene for 1 sec, the bright area gets washed out but the dark area is exposed just right. is this normal? please help because I want to be sure before taking my camera to a service center.
Thanks you very much.

[tim]

Post a couple of sample photos, pref straight from the camera, or even better put RAW files on a web server and post the links.

[PacAce]

hey guys before we go on super technical details, please help me. The main reason I post this is because I'm suspecting my XT has a problem in its dynamic range, I think it's to narrow. normally I use partial metering since that's the only mode available on XT closest to spot metering (I think). I notice that bright areas on a scene gets washed out easily even if the camera measures only about 2 2/3 stops of dynamic range. for example, in Av mode partial metering, i get a 1sec for the dark areas and 1/8sec on bright areas. that's just about 3 stops right, when I set the camera to expose the scene for 1 sec, the bright area gets washed out but the dark area is exposed just right. is this normal? please help because I want to be sure before taking my camera to a service center.
Thanks you very much.
Yes, that's normal. 1 sec. to 1/8 sec. is a 3-stop difference, putting anything that should have been exposed at 1/8 on the very bright part of the "spectrum" (i.e. looks washed out) when exposed at 1 sec. :)

If the dark area is supposed to be very dark, then the exposure for that part should be given less exposure than recommended by the camera because the meter always wants to make everything mid-gray in tone. Like-wise, the bright areas needs to be given more exposure then recommended for the same reason. If you had taken the shot with, say, 1/2 or 1/4 sec, then both the dark area and the bright area would probably have come out looking more properly exposed.

[randalcandari]

Yes, that's normal. 1 sec. to 1/8 sec. is a 3-stop difference, putting anything that should have been exposed at 1/8 on the very bright part of the "spectrum" (i.e. looks washed out) when exposed at 1 sec. :)

If the dark area is supposed to be very dark, then the exposure for that part should be given less exposure than recommended by the camera because the meter always wants to make everything mid-gray in tone. Like-wise, the bright areas needs to be given more exposure then recommended for the same reason. If you had taken the shot with, say, 1/2 or 1/4 sec, then both the dark area and the bright area would probably have come out looking more properly exposed.

Yeah, I tried stepping the exposure compensation down to -1 and the image looks much better. thanks everyone! still much more to learn

-randal