Ad
Ad
Ad
Pages: [1] 2 »   Bottom of Page
Print
Author Topic: Out of Focus Highlights not smooth?  (Read 7183 times)
OpticalMedia
Newbie
*
Offline Offline

Posts: 45


« on: December 07, 2009, 09:24:30 AM »
ReplyReply

I'm not too sure if this is an anomaly or if it is normal with my setup? 5d mark2 with 24-105 L lens. The out of focus highlights seem very badly transitioned and the one has a unattractive blue hue around. This may be expected as the sky is blue but surly the transition should be smoother? check at the sunset shot the highlight is blown as expected however is it normal to transition so unattractively?  All images where shot and processed with everthing zeroed in adobe camera raw. The images are basically flatted unprocessed jpegs.
« Last Edit: December 07, 2009, 09:28:28 AM by OpticalMedia » Logged
Jonathan Wienke
Sr. Member
****
Offline Offline

Posts: 5759



WWW
« Reply #1 on: December 07, 2009, 10:12:28 AM »
ReplyReply

Quote from: OpticalMedia
I'm not too sure if this is an anomaly or if it is normal with my setup? 5d mark2 with 24-105 L lens. The out of focus highlights seem very badly transitioned and the one has a unattractive blue hue around. This may be expected as the sky is blue but surly the transition should be smoother? check at the sunset shot the highlight is blown as expected however is it normal to transition so unattractively?  All images where shot and processed with everthing zeroed in adobe camera raw. The images are basically flatted unprocessed jpegs.

It's a clipping problem. Read this article for an explanation of the cause of the problem and what you can do to fix it.
Logged

Panopeeper
Sr. Member
****
Offline Offline

Posts: 1805


« Reply #2 on: December 07, 2009, 10:19:30 AM »
ReplyReply

ISO 50

The 5D2, like all other Canon DSLRs, does not have any ISO 50. If you select 50, the metering occurs for 50 but the shot is made with 100. When converting, the raw processor, at least DPP and LR/ACR compensate for the extra exposure by applying -1 EV adjustment - but that does not gain back the blown details. Thus your shots were done with +1 EV exposure bias.

Never use ISO 50 when shooting raw, and very cautiously with JPEG. (If we are at it: and never use 1/3 stop ISOs with raw.)
Logged

Gabor
OldRoy
Sr. Member
****
Offline Offline

Posts: 427


WWW
« Reply #3 on: December 07, 2009, 12:03:52 PM »
ReplyReply

Quote from: Panopeeper
ISO 50

.... (If we are at it: and never use 1/3 stop ISOs with raw.)
Hi PP.
Pardon my ignorance, but I don't understand this proviso.
Roy
Logged
Panopeeper
Sr. Member
****
Offline Offline

Posts: 1805


« Reply #4 on: December 07, 2009, 01:13:01 PM »
ReplyReply

Quote from: OldRoy
I don't understand this proviso.
I assume the question relates to the 1/3 ISO steps.

Only the 1Dxxx models have analog gain for the 1/3 EV ISO steps; the others are using the next lower respectively the next higher ISO and numerically manipulate the resulting data. The effect is, that the full stop +1/3 steps are using too low exposure (for example metered for ISO 250 but shot with ISO 200), but the result gets numerically adjusted and the top 1/3 EV clipped. On the other hand, the full step - 1/3 EV ISO steps are "overmetered" by 1/3 EV. For example ISO 320 is metered for 320, but shot with 400.

For a detailed explanation see ISO Setting with Canon 7D. It is correct regarding the 5D2 as well, except for the chapter The speciality of ISO 100 and 125.
Logged

Gabor
OpticalMedia
Newbie
*
Offline Offline

Posts: 45


« Reply #5 on: December 07, 2009, 01:26:06 PM »
ReplyReply

Quote from: Panopeeper
ISO 50

The 5D2, like all other Canon DSLRs, does not have any ISO 50. If you select 50, the metering occurs for 50 but the shot is made with 100. When converting, the raw processor, at least DPP and LR/ACR compensate for the extra exposure by applying -1 EV adjustment - but that does not gain back the blown details. Thus your shots were done with +1 EV exposure bias.

Never use ISO 50 when shooting raw, and very cautiously with JPEG. (If we are at it: and never use 1/3 stop ISOs with raw.)


So you saying the reason the clipping looks so unattractive etc is because I used iso L (which is 50 on my 5d mark 2 correct?) Why would canon have put this option in if it does this? Are you saying that if I shot iso 100 - then this problem I am referring to would not have occurred? And finally to clarify things if I shot at iso 125 for example would that mean I would get the same result as iso 50 (the problem im currently experiencing?)

best

Mike
Logged
Panopeeper
Sr. Member
****
Offline Offline

Posts: 1805


« Reply #6 on: December 07, 2009, 01:58:57 PM »
ReplyReply

Quote from: OpticalMedia
So you saying the reason the clipping looks so unattractive etc is because I used iso L (which is 50 on my 5d mark 2 correct?)
The shots you have posted were horrendeously overexplosed. Shooting with ISO 100 reduces the exposure by one stop (if you use the camera's metering). I have no way to say, if the shot was overexposed only by one stop - it is possible, that even the one stop reduction will not be enough and you have to apply a negative bias.

Quote
Why would canon have put this option in if it does this?
For those, who believe that they need it, no matter how. For example those shooting water falls want to increase the shutter time, no matter at what cost. Now, you can increase it by a +1 EV exposure bias, or by manual exposure, but if you are recording JPEG, the result will be too bright. USing ISO 50 instructs the camera to adjust the data before generating the JPEG. The same is happening with the raw converters: they do apply a -1 EV "Exposure" (with LR/ACR) or "Brightness" (with DPP) correction, thereby making up for the extra high exposure - but that can not help with clipping in the raw data.

Quote
Are you saying that if I shot iso 100 - then this problem I am referring to would not have occurred?
It would have been one stop less; perhaps that would have been enough to prevent clipping, and most probably it would have been enough to prevent leaving, which occured, if I see it right.

Quote
And finally to clarify things if I shot at iso 125 for example would that mean I would get the same result as iso 50 (the problem im currently experiencing?)
No;

- ISO 125 would have dictated an exposure 4/3 EV lower than what it was required with ISO 50,

- the shot would have been made with ISO 100,

- the resulting data would have been increased before recording the raw data, possibly causing a clipping by 1/3 EV.
Logged

Gabor
Guillermo Luijk
Sr. Member
****
Offline Offline

Posts: 1275



WWW
« Reply #7 on: December 07, 2009, 01:59:11 PM »
ReplyReply

Quote from: OpticalMedia
So you saying the reason the clipping looks so unattractive etc is because I used iso L (which is 50 on my 5d mark 2 correct?) Why would canon have put this option in if it does this? Are you saying that if I shot iso 100 - then this problem I am referring to would not have occurred? And finally to clarify things if I shot at iso 125 for example would that mean I would get the same result as iso 50 (the problem im currently experiencing?)

Any clipped image can display unattractive colour transitions in the partially saturated highlights, this is a disadvantage of digital, but if you set ISO50 these artifacts will be even more visible. The reason is that by setting ISO50 the camera works internally at ISO100 (therefore blowing more information than a real ISO50 would have), but afterwards a tone curve with a -1EV reduction is applied, and this -1EV compensation magnifies the artifacts found in the clipped highlights.

Your image would have looked better in the highlights by shooting at ISO100, and half exposure time. The price would have been some extra noise in the shadows.

Regards
Logged

OpticalMedia
Newbie
*
Offline Offline

Posts: 45


« Reply #8 on: December 07, 2009, 03:09:51 PM »
ReplyReply

Quote from: GLuijk
Any clipped image can display unattractive colour transitions in the partially saturated highlights, this is a disadvantage of digital, but if you set ISO50 these artifacts will be even more visible. The reason is that by setting ISO50 the camera works internally at ISO100 (therefore blowing more information than a real ISO50 would have), but afterwards a tone curve with a -1EV reduction is applied, and this -1EV compensation magnifies the artifacts found in the clipped highlights.

Your image would have looked better in the highlights by shooting at ISO100, and half exposure time. The price would have been some extra noise in the shadows.

Regards

Ran some of my own tests. Mirror lockup and manually focused so that highlight was out of focus. It is clear now that at ISO 50(or L) the out of focus highlight transitions look terrible! ISO 100 much better ! I have no idea who would use ISO 50??!! The third Image is ISO 125 the fourth ISO 160 and the 5th ISO 200. Images are all 100% grabs. No real world difference between the rest of the images after ISO 50...So i am still not 100%sure what panopeeper was on about with the rest of the ISO expansion options in between? Shed some light please, thanks a lot!
Logged
Panopeeper
Sr. Member
****
Offline Offline

Posts: 1805


« Reply #9 on: December 07, 2009, 04:36:49 PM »
ReplyReply

Quote from: OpticalMedia
No real world difference between the rest of the images after ISO 50...So i am still not 100%sure what panopeeper was on about with the rest of the ISO expansion options in between?
The difference between these is is not much, because the difference in the exposure is only 1/3 EV between the neighboring shots, and the camera adjusted the raw data to look like it would have been shot with the corresponding ISO gain (although this is not possible in all cases). Furthermore, the raw processors too apply their special twists. For example ACR applies +0.4 EV Exposure adjustment (clandestine, not visible on the slider) to all 5D2 shots with full-stop ISO (except ISO 50, which is adjusted by -0.6 EV); however, the full top -1/3 EV ISO shots are adjusted only by +0.08 EV. Thus the result of the conversion is pretty far from the raw data.

There is no disadvantage when using the -1/3 EV ISO steps, except that the metering is 1/3 EV too high, that can cause clipping. The +1/3 EV steps are undermetered and adjusted afterwards; that adjustment may cause clipping.

Anyway, these shots are not useful to demonstrate the differences; a well-defined scenery with clearly recognizable clipping can show the issues better, like in the 7D demo I linked to above. The raw files of these shots might be more useful to show, what happened.
Logged

Gabor
Eric Myrvaagnes
Sr. Member
****
Offline Offline

Posts: 7866



WWW
« Reply #10 on: December 07, 2009, 06:23:33 PM »
ReplyReply

Gabor ("panopeeper") is right.

The simplest way to think of it, for Canon owners anyway, is this: The only "real" ISO settings are 100, 200, 400, etc. Any lower or in-between settings are fake (they acrtually use the nearest "real" ISO number), andmay cause problems. I just never use them, and it doesn't cramp my photography in any meaningful way.

Eric

Logged

-Eric Myrvaagnes

http://myrvaagnes.com  Visit my website. New images each season.
Chris Pollock
Full Member
***
Offline Offline

Posts: 213


« Reply #11 on: December 08, 2009, 05:16:27 AM »
ReplyReply

Quote from: EricM
Gabor ("panopeeper") is right.

The simplest way to think of it, for Canon owners anyway, is this: The only "real" ISO settings are 100, 200, 400, etc. Any lower or in-between settings are fake (they acrtually use the nearest "real" ISO number), andmay cause problems. I just never use them, and it doesn't cramp my photography in any meaningful way.

Eric
This is very unsettling news. I was aware that ISO 50 is fake, and therefore never use it, but I had assumed that the 1/3 stop ISO values were real, and consequently used them quite often. I'll avoid doing so from now on.

I think it's unethical to add fake features like this, which do nothing more than cause potential problems. If I want to deliberately over or under expose my shot and then adjust it digitally I can do that myself thank you very much - I don't want the camera doing it behind my back. If Canon insist on adding such features, they should at least put a prominent warning in the manual, so that people will know not to use them.
Logged
Guillermo Luijk
Sr. Member
****
Offline Offline

Posts: 1275



WWW
« Reply #12 on: December 08, 2009, 05:33:55 AM »
ReplyReply

Quote from: Chris Pollock
I think it's unethical to add fake features like this, which do nothing more than cause potential problems. If I want to deliberately over or under expose my shot and then adjust it digitally I can do that myself thank you very much - I don't want the camera doing it behind my back. If Canon insist on adding such features, they should at least put a prominent warning in the manual, so that people will know not to use them.
You are making the mistake here of thinking as a RAW shooter, when unfortunately cameras today are still JPEG oriented (see some of my complaints about this here).

When shooting JPEG fake ISOs are useful because you can't (or you shouldn't) post process your images, and therefore a finished image has to come out right from the camera. Think of the following 2 situations where fake ISOs can be useful to a JPEG shooter:

- You want to do a silk effect in the water of a river. The sunlight is so bright that at ISO100 and even with the lens stopped to f/22, the exposure time needed for your silk effect means you are globally overexposing your image. But by setting ISO50 and the same aperture and shutter, you get a proper global exposure. You won't recover any clipped highlights (since the RAW data are the same as shooting at ISO100 with the same aperture and shutter), but thanks to the camera ISO50 post processing you will get proper exposure for most of your JPEG that resulted overexposed shooting at ISO100.

- You are street shooting at night, in very low light conditions. Even at max aperture (f/2.Cool and the longer exposure time your hands can afford, you get very dark JPEG images at the max real ISO (ISO1600/ISO3200 on most cameras, including the 5D2 BTW). Fake high ISO values such as ISO6400 or ISO12800 can provide you a properly exposed image. Of course they will be noisy, but it's better a noisy JPEG than no JPEG at all.

Of course the same philosophy that is valid for JPEG shooters, can be applied to RAW shooters who do not know/do not care about the inner workings of their cameras and RAW developers, and feel happy shooting at any ISO value provided by their camera.

Regards.
« Last Edit: December 08, 2009, 06:02:26 AM by GLuijk » Logged

Chris Pollock
Full Member
***
Offline Offline

Posts: 213


« Reply #13 on: December 08, 2009, 06:10:15 AM »
ReplyReply

Quite frankly I don't think anyone who cares enough about photography to spend thousands of dollars on a camara should be shooting JPEG, but that's another story.

I have nothing against Canon implementing fake features for the benefit of JPEG shooters, but they really ought to either disable such features in raw mode, or add a warning that you should only use them if you don't much care about quality.

I've done a little bit of internet research, and it seems that there's a lot of disagreement about whether the 1/3 stop ISO values on the 5D and 5D II are fake or not. Some people also think that ISO 6400 on the 5D II is fake, although it's a moot point for me - I never shoot above ISO 3200 due to excessive noise. I have no idea which information is reliable, so to be safe I'll avoid the 1/3 stop ISO values until such time as I find reliable information.
Logged
Guillermo Luijk
Sr. Member
****
Offline Offline

Posts: 1275



WWW
« Reply #14 on: December 08, 2009, 06:25:00 AM »
ReplyReply

Quote from: Chris Pollock
Quite frankly I don't think anyone who cares enough about photography to spend thousands of dollars on a camara should be shooting JPEG, but that's another story.

I have nothing against Canon implementing fake features for the benefit of JPEG shooters, but they really ought to either disable such features in raw mode, or add a warning that you should only use them if you don't much care about quality.

I've done a little bit of internet research, and it seems that there's a lot of disagreement about whether the 1/3 stop ISO values on the 5D and 5D II are fake or not. Some people also think that ISO 6400 on the 5D II is fake, although it's a moot point for me - I never shoot above ISO 3200 due to excessive noise. I have no idea which information is reliable, so to be safe I'll avoid the 1/3 stop ISO values until such time as I find reliable information.
If you want to know what information is reliable, produce your own information. There is no mistery in extracting the RAW data and analysing it.

ISO6400 on the 5D2 is fake. It is not a matter of thinking it or not, just a matter of analysing the RAW files and seeing it; just look at these RAW histograms. Regarding the 1/3 stop ISOs, just read Gabor's (Panopeeper) findings.

Do you use your camera's histogram? that histogram is related to the embedded JPEG file, so it can be considered a fake histogram if you are shooting RAW (it displays information that has been clipped, mainly because of white balance, so it is not reliable to check how good your RAW data are. Just take a look: Camera's histogram reliable to the RAW data, Specific WB for realistic camera JPEG). People taking care about fake ISOs should also be concerned about a histogram that is lying to them, and seeing a bit farer, about how camera manufacturers are still making JPEG-oriented cameras, no matter how much they cost.

Regards
« Last Edit: December 08, 2009, 06:26:47 AM by GLuijk » Logged

Chris Pollock
Full Member
***
Offline Offline

Posts: 213


« Reply #15 on: December 08, 2009, 06:38:11 AM »
ReplyReply

Quote from: GLuijk
Do you use your camera's histogram? that histogram is related to the embedded JPEG file, so it can be considered a fake histogram if you are shooting RAW (it displays information that has been clipped, mainly because of white balance, so it is not reliable to check how good your RAW data are. Just take a look: Camera's histogram reliable to the RAW data, Specific WB for realistic camera JPEG). People taking care about fake ISOs should also be concerned about a histogram that is lying to them, and seeing a bit farer, about how camera manufacturers are still making JPEG-oriented cameras, no matter how much they cost.
Regards
Yes, I was aware of that, and it annoys me greatly. I've commented on it a few times on other threads. For that reason I tend to bracket my landscape shots 1/3 of a stop either way, and examine them in ACR before deciding which one to keep. A proper raw histogram would be extremely useful, and would be easier to implement than the JPEG-oriented crap we have now. Having said that, lying about the histogram doesn't make it OK to lie about the ISO too. I don't think we really disagree about anything here.
Logged
Chris Pollock
Full Member
***
Offline Offline

Posts: 213


« Reply #16 on: December 08, 2009, 06:45:30 AM »
ReplyReply

Quote from: GLuijk
If you want to know what information is reliable, produce your own information. There is no mistery in extracting the RAW data and analysing it.
I think I'll do that when I have time. It didn't occur to me to do so until now, because I always assumed that all the ISO values other than the "expanded" ones were real. Now I have evidence to the contrary.
Logged
thierrylegros396
Sr. Member
****
Offline Offline

Posts: 647


« Reply #17 on: December 09, 2009, 02:59:28 AM »
ReplyReply

And what about other brands like Nikon, Olympus, Sony ?!

Are 1/3 ISO steps fake ISO too ?

This is a very interresting thread and I think that it could turn into a more complete Tutorial or informative article because a lot of people ignore that !

Yes most of us know that extreme ISO are fake, but not 1/3 ISO steps.


Have a Nice Day !

Thierry
Logged
PierreVandevenne
Sr. Member
****
Offline Offline

Posts: 510


WWW
« Reply #18 on: December 09, 2009, 10:33:47 AM »
ReplyReply

Quote from: thierrylegros396
Yes most of us know that extreme ISO are fake, but not 1/3 ISO steps.

Well, generally speaking, all ISO are fake in a digital camera. It's just a convenience, normalization. This is why 100 ISO doesn't really mean exactly the same thing in a camera or another. At the root, we have a photon counter: it doesn't really matter much if we multiply/divide at the AD converter, in the camera, or in software post-processing as long as the signal sits nicely between the points where noise becomes an issue and the point where horrible things such as dramatically non linear response, blooming, clipping occurs...

But of course, in practice, it helps a lot to have a camera normalized to the conventions we are familiar with and one that doesn't output a trash signal by default...
Logged
Panopeeper
Sr. Member
****
Offline Offline

Posts: 1805


« Reply #19 on: December 09, 2009, 12:43:00 PM »
ReplyReply

Quote from: thierrylegros396
And what about other brands like Nikon, Olympus, Sony ?!

Are 1/3 ISO steps fake ISO too ?
The 1Dxxx models do have analog ISO gain for the 1/3 steps as well.

Nikons, at least the D300 and above have analog gain for the 1/3 steps; I don't have such raw files from the lower and older models.

The Sony A700 seems to have analog gain, but I am not sure about that, because the raw data is much less raw than Canon's.

I don't have such raw files from most other cameras; I looked at the Olympus E3, that seems to be analog.
Logged

Gabor
Pages: [1] 2 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad