ISO 4000....my jaw dropped at the quality. Have fun with your new toy! []
ISO 4000....my jaw dropped at the quality. Have fun with your new toy! [
If it were me I would be careful about pushing the ISO so high. It's nice to see the camera can do it without noise getting out of hand but with the exception of the last two shots all others are a bit over exposed. Notice the inconsistant color of the church interior from shot to shot. #5 (vertical) is probably the closest to reality. High lights of the stained glass and doorway are blown out and I'll bet the histogram confirms this. I have shot many many pictures in many churches so setting a custom white balance and getting consistant exposure is key.
I must say that I'm not entirely with you on your analysis Tom.
Originally Posted by Tom Wertman
Agreed on this part, and Alex also look out for the tweener ISO's. ISO 2000 for example looks lower than ISO 3200 and so you might expect lower noise levels, but that isn't true. That only counts for true ISO values: 100-200-400-800-1600-3200-6400. ISO 2000 for example is a picture taken with ISO 3200 and processed 2/3 stop underexposed in camera-processing. It might not technically precisely work like this, but the idea is sound. So you get the idea [] Therefor I personally have set my ISO to only use full stop numbers.
I'm not sure what your idea is of over/underexposure, but the lastoneactually seems underexposed to me. I can't imagine a shot like that one exposed properly with those settings: 1/250 f11 ISO500. With those settings and lighting I would assume this shot is taken early or lately during the day, but we don't see the actual level of light that was available at that point. Mostly because for having those settings, a pretty decent amount of light is needed to expose naturally, which we don't really get to see. I do like the shot, but I don't think it is "naturally" exposed. But I guess this shot was intended this way and proper exposure is very subjective?
I agree that custom white-balance or post-processing white balance adjustment would be a good idea.
I don't think this is an honest challenge since the camera would never be able to capture the actual dynamic range that is available there. You must lose some detail in either the darker parts or the lighter parts of the picture. If I were to shoot my own room or a church or whatever else indoor with a window viewing outside and I would expose for the interior it would seem obvious to me that the windows and everything else of the exterior would be blown out. The camera just can't handle such a wide dynamic range. HDR would cover this part pretty good I think. But with those shutterspeeds and without a tripod I wouldn't take my chances [A]
I guess my idea of proper exposure is different than yours, so just look at my idea as an opinion and not the absolute truth.
If you want to, I'm really interest in a shot of the interior of a church or something with consistent exposure and white balance so I would really like to see an example from you. Perhaps it will change my opinion.
Jan
I'm not sure this is correct, Jan. Photon noise is a function of sensitivity and the amount of light hitting the camera. When you set your camera to ISO 2000, it uses a slower shutter speed and thus lets more light in than it would have if you had used ISO 3200.
The problem with fake ISO's isn't that you get too much noise, but rather, you lose range (see below)
The camera also "overexposes" the picture by 2/3 of a stop as compared to an ISO 3200 shot. Thus you lose highlight headroom.
Congrats on your purchase, Alex.
I really want to go full frame. Just waiting to graduate. After 8 years of post high-school studies, I think I deserve it. Ha!
Keep shooting! Then post.
Dave.
5D mark III, 50D, 17-40 f4L, 24-70 f2.8L, 70-200 f4L IS, 28 f1.8, 50 f1.8, 85 f1.8, 100 f2.8 Macro
Thanks alot, Dave.
I say go for it. You have worked hard, and you certainly deserve it!
Now I feel weird about this whole thing - I'm in my final year of high school. Just saying. [H]
Here's one more from the other day... I grabbed this shot as I rode past the scene of a car accident (I believe everyone was OK), where some firefighters were leaning on the guard rail and seemed to be exhausted after a long day of hard work.
Check it out:
5D Mark II, EF 70-200mm f/2.8 L IS, 1/1000 sec., f/2.8, ISO 200 @ 200mm. Minor sharpening and lowered saturation in Lightroom 2.0. I kind of like the fact that you can see the "TORONTO" embroidered into the back of their jackets - Sort of cool.
C&C welcome.
[img]/cfs-file.ashx/__key/CommunityServer.Components.UserFiles/00.00.00.26.94/GardinerCrash_2D00_1.jpg[/img]
To be clear, here is the whole story behind ISO 2000:
- Sets the camera to ISO 1600 (but displays "2000")
- Adds a hidden -1/3 stop EC to the autoexposure meter.
- After the photo is taken, it increases brightness of the raw file by 1/3 stop (with a linear digital push)
<div>The net result is that you lose 1/3 stop of highlight headroom, just as you said. You can simulate ISO 2000 by setting ISO 1600, -1/3 EC, and then increasing brightness in post. The only difference is that when you do it yourself, you can increase brightness in a way that doesn't clip highlights.</div>
<div></div>
<div>It's the same story with all the "plus 1/3" ISO settings: 125, 250, 500, 1000, 2000, 4000.</div>
So it is the opposite direction from what Jan and I were saying...
But then, don't you lose 1/3 stop in the darks instead of in the highlights? Doesn't doing a linear digital push leave you with no data at the low end of the raw file? Or do you mean you lose 1/3 stop of highlight headroom as compared to iso1600 underexposed by 1/3 stop (but not as compared to a "correctly" exposed picture?
Anyway, it bad.
If canon would just leave the raw alone, the world would be a better place.
Normally it would mean you lose 1/3 in the highlights *and* some in the shadows, but in this case it's only highlights because two of Canon's flaws cancel each other out in the shadows. I'll explain:
If Canon allowed their engineers to design the camera, then they would choose the correct design design where raw files have a bit depth appropriate for the noise level. In that case, mucking about with the raw data like this would have negative consequences in the deep shadows by causing quantization error. Such posterization would probably cause the loss of even more than 1/3 stop, but that's in shadows much darker and deeper than most photographers ever go, so in practice most people wouldn't notice even if it caused the loss of a full stop. Of course, if you were going to choose a sensible design you wouldn't allow this sort of mucking about in the first place.
In any case, the reality of the situation is that Canon cameras are clearly not designed by the engineering department, but the Marketing department. So instead of a sensible raw file design (11 or 12 bits), they use 14-bits. At normal ISO settings, all those extra bits go to waste. But on the 1/3 stop ISO settings, where Canon mucks about with the raw data for no reason, the extra bits prevent quantization error. So the two flaws cancel each other out. Of course it would be better if we had neither flaw, but oh well.
If Canon had a *really* good design, like Nikon's compressed NEF, then mucking with the raw would have *really* bad effects, because then it would affect *all* tonal levels instead of just shadows. That's precisely what Nikon does and there are some NEF files floating around that show the damage (posterized skies, etc.). Nikon seems to combine brilliance with stupidity, while Canon is more of mediocrity with mediocrity. At least we have variety. [
Technically, no. That's because raw files don't go from white to black, they go from white to noise. You just put black wherever you want (most people choose a spot that is far away from the noise). Because of this, the net effect is that it basically just decreases the resolution that it can quantize with throughout all the tonal levels. In Canon cameras, photon shot noise will completely hide the loss in resolution at all tonal levels except for the deepest shadows, so that's the only place where the negative effect could be seen (if it had a more appropriate bit depth).
On these ISO settings, Canon literally takes the top 1/3 stop of values and deletes them.
Agreed. And if the engineers could somehow find a way to trick the Marketing department into thinking they were designing the cameras one way, but secretly do it the correct way. Oh, I can dream...
Posting Permissions
Reply With Quote