| Author | Thread |
|
|
02/11/2007 02:47:30 PM · #1 |
OK, so I thought I understood but I guess I don't. This is what I thought.
f-stop has to be the same for all the shots, therefore you can only shoot on Manual (no AV and no TV). No auto focus and no auto white balance.
So I went out and shot on manual about 20 exposures, some came out completly underexposed (entire frame black), so I can't use those. So that doesn't make sense. The only way it worked was when I shot on Av and changed exposure compensation in full increments but that only gave me 4 frames.
I don't get it. Can someone explain? |
|
|
|
02/11/2007 03:04:40 PM · #2 |
On the 20D, you can set up exposure bracketing. If you then use burst mode, the camera will take three pictures, one 'normally' exposed, one underexposed and one overexposed. Av mode works well for this, as the aperture value is fixed and the camera varies the shutter speed. This is quite useful, too, if you don't have your tripod with you. Plus and minus two stops is a pretty good place to start, I think, for the bracketing. The thing to watch out for, though, is whether the shutter speed on the over exposed picture will be too slow.
|
|
|
|
02/11/2007 03:25:37 PM · #3 |
Originally posted by Mr_Pants:
On the 20D, you can set up exposure bracketing. If you then use burst mode, the camera will take three pictures, one 'normally' exposed, one underexposed and one overexposed. Av mode works well for this, as the aperture value is fixed and the camera varies the shutter speed. This is quite useful, too, if you don't have your tripod with you. Plus and minus two stops is a pretty good place to start, I think, for the bracketing. The thing to watch out for, though, is whether the shutter speed on the over exposed picture will be too slow. |
Thanks. But what if I wanted to take more than 3 exposures? |
|
|
|
02/11/2007 03:46:28 PM · #4 |
To do this in manual mode you need to choose the aperture then keep an eye on the exposure bar in the viewfinder while choosing your shutter speed.
If you want to do more than -2 or + 2 then just count the number of adjustments below or above the lowest/highest setting in the viewfinder. Since three changes in one direction will give you one stop it's quite easy to do.
example:
-2 - - -1 - - 0 - - +1 - - +2
*
When the marker at this (*) point is not flashing then the camera is showing you that the setting is 2 stops under exposed. If you want three, dial it in. The (*) will then be flashing but that's ok since you know what you want to set. |
|
|
|
02/11/2007 04:33:21 PM · #5 |
| For those that have done a lot of HDR work, do you think there is much advantage to using 5 images instead of 3? |
|
|
|
02/11/2007 09:01:59 PM · #6 |
Originally posted by cpanaioti:
To do this in manual mode you need to choose the aperture then keep an eye on the exposure bar in the viewfinder while choosing your shutter speed.
If you want to do more than -2 or + 2 then just count the number of adjustments below or above the lowest/highest setting in the viewfinder. Since three changes in one direction will give you one stop it's quite easy to do.
example:
-2 - - -1 - - 0 - - +1 - - +2
*
When the marker at this (*) point is not flashing then the camera is showing you that the setting is 2 stops under exposed. If you want three, dial it in. The (*) will then be flashing but that's ok since you know what you want to set. |
thanks. when you say 3 changes in one direction give you one stop, I assume you mean when the exposure compensation is set to 1/3 increments, right?
How far would you go though? I wanted to make about 10 exposures but of course that may not be possible in all instances.
Also, when you shoot in RAW, do you process them with all the same settings? |
|
|
|
02/11/2007 09:24:19 PM · #7 |
|
|
|
02/12/2007 02:44:08 PM · #8 |
|
|
|
02/12/2007 03:15:15 PM · #9 |
Originally posted by maggieddd:
OK, so I thought I understood but I guess I don't. This is what I thought.
f-stop has to be the same for all the shots, therefore you can only shoot on Manual (no AV and no TV). No auto focus and no auto white balance.
So I went out and shot on manual about 20 exposures, some came out completly underexposed (entire frame black), so I can't use those. So that doesn't make sense. The only way it worked was when I shot on Av and changed exposure compensation in full increments but that only gave me 4 frames.
I don't get it. Can someone explain? |
First off, nobody really wants to shoot for HDR - High Definition Range. That is when the bright areas of a composition are MUCH brighter than the darker areas of a composition.
You shoot for HDR(High Definition Range) only when you MUST shoot in an environment where there is a large lighting difference between bright and dark items in a composition. Traditionally, photographers either avoid that or reject photography under those conditions. You should to.
But suppose, for example, that you want to shoot a sunset with the sun still above the horizon but there is great color and/or cloud detail far away from the sun in darker areas and you want to bring that out detail in a composition in that area without overexposing the sun. Then you might consider HDR.
In that case you would take a series of pictures all with different exposures either changing just the shutter speed or the f/stop depending on the situation. For a sunset that would be the shutter speed if everything in the composition were close to infinite distance. You might vary f/stop if that were not the case. Most people use HDR to try to overcome that.
Sometimes DPC photographers try to 'save' poorly lighted images using HDR and/or apply it where other techniques might work just as well or better.
Message edited by author 2007-02-12 15:31:58.
|
|
|
|
02/12/2007 03:19:59 PM · #10 |
Steve, a lot of people actually do shoot for the high dynamic range style, all the time - even in situations that don't really warrant it. The link to Trey's site at //stuckincustoms.com is a good example. He uses HDR & tone mapping for everything, to create his style - even in plenty of cases were it isn't needed.
|
|
|
|
02/12/2007 03:30:34 PM · #11 |
Originally posted by Megatherian:
For those that have done a lot of HDR work, do you think there is much advantage to using 5 images instead of 3? |
It all depends on the dynamic range in the scene - and how much you bracket it for each of those stops. It probably isn't worth bracketing in less than 1 stop increments, so if you capture the full range of what you are interested in, in those 3 frames, there isn't much point in adding more detail in.
If you assume a typical digital shot captures around 5 stops of dynamic range, if you under expose by one stop, you pick up one extra stop at the low end, same with the high end, so you've now got a 7 stop range result.
If you bracket a further +-1 for a 5 stop shot, you get 9 stops of dynamic range in the final result - if you need that much, then you need to bracket that much - and so it goes on.
You can work it out by metering your highlights and shadows that you want to still have detail in and expose for the number of stops difference between those.
|
|
|
|
02/12/2007 03:32:45 PM · #12 |
Originally posted by Gordon:
Steve, a lot of people actually do shoot for the high dynamic range style, all the time - even in situations that don't really warrant it. The link to Trey's site at //stuckincustoms.com is a good example. He uses HDR & tone mapping for everything, to create his style - even in plenty of cases were it isn't needed. |
great photos on his page, and he's currently shooting in iceland!!!
|
|
|
|
02/12/2007 03:33:12 PM · #13 |
Originally posted by Megatherian:
For those that have done a lot of HDR work, do you think there is much advantage to using 5 images instead of 3? |
I can tell you what's an absolute DISadvantage: shooting/merging a series where the darkest exposure actually underexposes the bright areas, and the lightest exposure overexposes the dark areas. When you do an HDR merge that includes images on the extremes that are too bright or too dark for the darks/brights respectively, you end up with a muddy merged image that doesn't work at all.
Hypothetically speaking, if the tonal range is two stops greater than your sensor can record effectively, then a plus-one and a minus-one are as far as you need to go. Adding a plus-two and a minus-two to the mix is counterproductive. On the other hand, if you have a tonal range that's four stops out-of-gamut, then the five 1-stop variations are needed/useful.
Another factor to consider is the subtlety of the merge; even if minus-one and plus-one are the caps of your range, you can arguably get a better result by exposing 5 exposures in 1/2 stop increments instead of three in 1-stop increments.
So it kind of depends on what you mean, exactly. I'd say, as a rule of thumb, that as long as your extreme images are correctly exposed for their relative targeted values, the more interim steps you dial in the better off you are. There is a point of diminishing returns where you can no longer see the difference, but I don't see any harm coming from that except extra processing time to generate the merge.
R.
|
|
|
|
02/12/2007 03:33:35 PM · #14 |
Originally posted by Gordon:
Steve, a lot of people actually do shoot for the high dynamic range style, all the time - even in situations that don't really warrant it. The link to Trey's site at //stuckincustoms.com is a good example. He uses HDR & tone mapping for everything, to create his style - even in plenty of cases were it isn't needed. |
Gordon, that does not surprise me and it does work for some folks. It still does not justify making it a standard practice.
|
|
|
|
02/12/2007 03:35:15 PM · #15 |
Originally posted by Gordon:
Originally posted by Megatherian:
For those that have done a lot of HDR work, do you think there is much advantage to using 5 images instead of 3? |
It all depends on the dynamic range in the scene - and how much you bracket it for each of those stops. It probably isn't worth bracketing in less than 1 stop increments, so if you capture the full range of what you are interested in, in those 3 frames, there isn't much point in adding more detail in.
If you assume a typical digital shot captures around 5 stops of dynamic range, if you under expose by one stop, you pick up one extra stop at the low end, same with the high end, so you've now got a 7 stop range result.
If you bracket a further +-1 for a 5 stop shot, you get 9 stops of dynamic range in the final result - if you need that much, then you need to bracket that much - and so it goes on.
You can work it out by metering your highlights and shadows that you want to still have detail in and expose for the number of stops difference between those. |
We're saying the same thing :-) But you made a boo-boo here. Underexposing by one stop gives you an extra stop of range on the bright end (the high end) and overexposing by a stop gives you a stop of extra range at the dark end (the low end).
R.
|
|
|
|
02/12/2007 03:37:30 PM · #16 |
Originally posted by Bear_Music:
Originally posted by Megatherian:
For those that have done a lot of HDR work, do you think there is much advantage to using 5 images instead of 3? |
I can tell you what's an absolute DISadvantage: shooting/merging a series where the darkest exposure actually underexposes the bright areas, and the lightest exposure overexposes the dark areas. When you do an HDR merge that includes images on the extremes that are too bright or too dark for the darks/brights respectively, you end up with a muddy merged image that doesn't work at all.
R. |
Have you tried masking your images prior to merging them ? Seems like most things, some manual intervention would lead to better results than expecting the tools to do it all automatically. I've seen a lot of manual HDR work done with masks that doesn't have this problem.
Another option might be to do a multi-pass merge, combine the lower exposures and create a darker HDR image for the shadows, combine the higher exposures, create a ligher HDR image for the highlights, then do a third pass, with the shadow HDR, a straight midtone exposure and your higher end HDR image and combine the results of all 3.
Don't tone map until the last stage, have the intermediate passes 'clean' and just for the dynamic range.
|
|
|
|
02/12/2007 03:39:31 PM · #17 |
Originally posted by Bear_Music:
Another factor to consider is the subtlety of the merge; even if minus-one and plus-one are the caps of your range, you can arguably get a better result by exposing 5 exposures in 1/2 stop increments instead of three in 1-stop increments. |
What's the argument in favour of that ?
|
|
|
|
02/12/2007 03:40:08 PM · #18 |
Originally posted by Bear_Music:
We're saying the same thing :-) But you made a boo-boo here. |
Yup.
|
|
|
|
02/12/2007 03:44:29 PM · #19 |
There's a great deal of confusion being fomented in DPC lately regarding what is and what isn't HDRI imaging. Technically, if the scene has a "normal" dynamic range, then you cannot have an HDRI version of it. What's happening here is that people are using one of the tools in the HDRI arsenal and applying it to "regular" pictures to produce a certain effect of "expanded local contrast". In Photomatix, the tool is "tone mapping". In CS2 it is "shadow/highlight adjustment".
I'd go so far as to say that nearly all of the images people are looking at and saying "OOH! So THAT'S what HDR can do!" are actually examples of local contrast enhancement, tone mapping, and not HDRI at all. "True" HDRI imaging is usually pretty-well integrated into the image; you have to know something about the subtleties of photographic reproduction to realize it's been used at all.
This is true HDRI: it's relatively natural looking.
This is tone mapping of a single exposure in basic: it's completely unnatural looking, albeit very dramatic.
Both finished in top 10 in their respective challenges. The first one actually ribboned.
R.
|
|
|
|
02/12/2007 03:46:06 PM · #20 |
Originally posted by Gordon:
Originally posted by Bear_Music:
Another factor to consider is the subtlety of the merge; even if minus-one and plus-one are the caps of your range, you can arguably get a better result by exposing 5 exposures in 1/2 stop increments instead of three in 1-stop increments. |
What's the argument in favour of that ? |
More subtlety of detail and local contrast in images with a lot of mid-tones.
R.
|
|
|
|
02/12/2007 03:47:03 PM · #21 |
Yup, I agree with Robert. You don't have to tone-map an image to death for it to have a high dynamic range. All though that over-processed look is very popular and prevalent in HDR images currently.
This is another HDR image, without much visible tone mapping.
|
|
|
|
02/12/2007 03:48:33 PM · #22 |
Originally posted by Bear_Music:
Originally posted by Gordon:
What's the argument in favour of that ? |
More subtlety of detail and local contrast in images with a lot of mid-tones. |
But wouldn't those mid-tones be well captured in the base exposure for the mid-tones (and the +-1 exposures around that midtone ?) Do you have any examples that show a quantifiable difference ? It seems to be redundant to capture them more than 3 times ?
Message edited by author 2007-02-12 15:48:52.
|
|
|
|
02/12/2007 03:54:35 PM · #23 |
Originally posted by Gordon:
Originally posted by Bear_Music:
Originally posted by Megatherian:
For those that have done a lot of HDR work, do you think there is much advantage to using 5 images instead of 3? |
I can tell you what's an absolute DISadvantage: shooting/merging a series where the darkest exposure actually underexposes the bright areas, and the lightest exposure overexposes the dark areas. When you do an HDR merge that includes images on the extremes that are too bright or too dark for the darks/brights respectively, you end up with a muddy merged image that doesn't work at all.
R. |
Have you tried masking your images prior to merging them ? Seems like most things, some manual intervention would lead to better results than expecting the tools to do it all automatically. I've seen a lot of manual HDR work done with masks that doesn't have this problem.
Another option might be to do a multi-pass merge, combine the lower exposures and create a darker HDR image for the shadows, combine the higher exposures, create a ligher HDR image for the highlights, then do a third pass, with the shadow HDR, a straight midtone exposure and your higher end HDR image and combine the results of all 3.
Don't tone map until the last stage, have the intermediate passes 'clean' and just for the dynamic range. |
I'm not sure I see any reason for merging an exposure that renders the brightest areas darker than I would want them to be. Or the darkest areas brighter than I would want them to be.
When you do an HDR merge, you are instructing the program where you want the limits to be. It then compresses those limits down to encapsulate everything else. End result, if you use too much range, is muddy whites and unnaturally bright shadow areas, with the midtones forced into a highly-compressed mode, which I don't see how, in general, is of any help at all. I am sure you can get some interesting effects this way, but we're sort of talking about the BASICS of the process here, aren't we?
Threads like this are happening because people don't understand the process very well and are seeking advice. In a nutshell, this is what HDRI is:
If, when you make an exposure that is correct for the bright areas the rest of the image goes too dark, and if you expose for the dark areas the rest of the image goes too bright, and if you expose in the middle the darks are blocked up and the brights are blown out, THEN you have the classic HDRI scenario. What you do is make an exposure that's right for the highlights, another that's right for the midtones, and another that's right for the dark areas (and possibly others filling the interim steps if the gaps between the first 3 are a stop or more) and then you merge all these images into a single composite image that gives you correct exposure in the middle and at both extremes.
(The above isn't instructing YOU, Gordon, 'cuz you know this. It's for the folks who are unclear on the topic.)
R.
|
|
|
|
02/12/2007 03:55:59 PM · #24 |
Originally posted by maggieddd:
Also, when you shoot in RAW, do you process them with all the same settings? |
The answer to that is, it depends (on the situation). By choosing smaller increments you can get more exposures.
If going from -2 to +2 in 1/3 increments you would have 13 exposures. Probably overkill, but you could do it.
|
|
|
|
02/12/2007 03:57:30 PM · #25 |
Originally posted by Gordon:
Originally posted by Bear_Music:
Originally posted by Gordon:
What's the argument in favour of that ? |
More subtlety of detail and local contrast in images with a lot of mid-tones. |
But wouldn't those mid-tones be well captured in the base exposure for the mid-tones (and the +-1 exposures around that midtone ?) Do you have any examples that show a quantifiable difference ? It seems to be redundant to capture them more than 3 times ? |
No, I don't. I'm talking theoretically. I have seen writings that say this is the case, but that's the next step in my experimentations with the process. I have seen some remarkable HDRI images comprised of like 11 exposures where it seemed to me that 5 would have been sufficient, but I haven't seen an example of the same thing done with 5. I'll make it a point of doing this next time I'm in an appropriate position to do so.
R.
|
|
