Ronald_H
Don't call me Ron
Recently I read in a few threads that people 'hated HDR'. Ok, maybe the HDR effect is now so popular that it gets boring, but being able to record a HDR image is great in itself right? I myself shot a few shots where the final result is way better than a 'single' shot, just because I could save shadows and highlights in a very contrasty scene.
Last weekend I went to a festival and made some happy snaps with my Olympus Mju-II. Not great art, just fun. Still, I got a nice shot of my friend Marjolijn:
That's quite a bit overexposed. But it's film so no worries (with a digital the shot would have been ruined).
To compensate I dialed down the gain of my Coolscan V:
Much better, but rather contrasty, no shadow detail left.
Then it hit me. Why not merge them as HDR in Photoshop? The final image is cleaned up and sharpened, but the tonal range was not changed:
Much, much better don't you think?
Last weekend I went to a festival and made some happy snaps with my Olympus Mju-II. Not great art, just fun. Still, I got a nice shot of my friend Marjolijn:
That's quite a bit overexposed. But it's film so no worries (with a digital the shot would have been ruined).
To compensate I dialed down the gain of my Coolscan V:
Much better, but rather contrasty, no shadow detail left.
Then it hit me. Why not merge them as HDR in Photoshop? The final image is cleaned up and sharpened, but the tonal range was not changed:
Much, much better don't you think?
Jamie123
Veteran
Could've easily been done in one scan. Next time just dial down the gain a bit less.
Keith
The best camera is one that still works!
I agree that the last image is much improved exposure wise by merging the two scans but it has lost a little sharpness and punch.
I prefer the tone and look of the second image to be truthful!
I prefer the tone and look of the second image to be truthful!
martin s
Well-known
Can't you just scan the full range and do a curve adjustment later? I don't see the problem.
martin
martin
craygc
Well-known
with negs film you're only dealing with a dynamic range of 2.0 at most so capturing what information is there should be reasonably easy for a good scanner single pass.
lorriman
Established
Not everyone knows about curves. Or they think you need CS or some kind of financial outlay. Or they think it's an advanced technique (which it is, but it's also absurdly easy).
Free curves plugin: smartcurves
Free pluggable viewer (quite nice in its own right): Irfanview
Curves guides
Free curves plugin: smartcurves
Free pluggable viewer (quite nice in its own right): Irfanview
Curves guides
nksyoon
Well-known
Pulling down highlights works well but an underexposed image would be difficult as pulling up shadows usually worsens grain/noise for me.
Ronald_H
Don't call me Ron
I do know about curves
Although the scanner will probably capture the DR in one pass, two passes gives a few more extra bits to get highlight separation. I cannot get the same results with curves, nor with highlight/shadow tools.
Even merging the two pics as seperate layers and playing with opacity does not give the same result (although it does come close).
There are many, many ways to get the DR of film (huge) to a LCD monitor, for this pic it works. For the next, who knows? But it is good to see how much you can still do with film.
Although the scanner will probably capture the DR in one pass, two passes gives a few more extra bits to get highlight separation. I cannot get the same results with curves, nor with highlight/shadow tools.
Even merging the two pics as seperate layers and playing with opacity does not give the same result (although it does come close).
There are many, many ways to get the DR of film (huge) to a LCD monitor, for this pic it works. For the next, who knows? But it is good to see how much you can still do with film.
Jamie123
Veteran
I'm actually more surprised by how much you can still NOT do with digital.
Anyways, my point was that even without curves you could have retained much of the shadow details if you had only dialed down the gain half as much.
Matus
Well-known
I don not think that using curves and merging two images with different exposure (be them two scans) is the same.
If you take the single scan and look at your highlights (am say that they are all there) - the separation is very week - so all your highlight sits close to the top of the curve (which is linear if you have not changed it yet). If you now apply some curves - say the typical "S" shaped you will increase the separation between highlight and shadow - especially in the medium grey parts (if the curve was more-less symmetric) However this is payed by compressing the very top (highlight) and very bottom (deep shadow) what makes them even more flat.
Applying only curves on the problematic part (could be done via coppying the part of image into separate layer) is very hard - as you want to make an S curve only within top most 5 - 10 percent of the range.
So if you want to expand those small details in your highlight detail (e.g. the shirt of the girl in the very first photo) you can first darken it (one possibility was the mentioned underexposure with scanner - what I would prefer to just darkening in PS) and if necessary apply some curves afterwards.
Now if the rest of the image should not be affected - you may then blend just the re-worked part of the image back to original (can be done with layer adjustment in PS for example) or take the more "lazy" approach and use the HDR. However in the latter case one looses the control of how this will be done.
My bottom line is: doing two separate scans and merging in some fashion
afterwards should help to improve the image quality in the very dark or very light areas. But I would guess it also depends a lot on the scanner and the operator.
****
Bellow you find your image after working on the original image on bright part of the subject a bit (I hope you do not mind) - I decreased the brightness with 'leves' and added a bit of curves afterwards. The rest of image remained untouched.
It looks indeed differently than your HDR (which looks better to me) but I just wanted to demonstrate the approach.
If you take the single scan and look at your highlights (am say that they are all there) - the separation is very week - so all your highlight sits close to the top of the curve (which is linear if you have not changed it yet). If you now apply some curves - say the typical "S" shaped you will increase the separation between highlight and shadow - especially in the medium grey parts (if the curve was more-less symmetric) However this is payed by compressing the very top (highlight) and very bottom (deep shadow) what makes them even more flat.
Applying only curves on the problematic part (could be done via coppying the part of image into separate layer) is very hard - as you want to make an S curve only within top most 5 - 10 percent of the range.
So if you want to expand those small details in your highlight detail (e.g. the shirt of the girl in the very first photo) you can first darken it (one possibility was the mentioned underexposure with scanner - what I would prefer to just darkening in PS) and if necessary apply some curves afterwards.
Now if the rest of the image should not be affected - you may then blend just the re-worked part of the image back to original (can be done with layer adjustment in PS for example) or take the more "lazy" approach and use the HDR. However in the latter case one looses the control of how this will be done.
My bottom line is: doing two separate scans and merging in some fashion
afterwards should help to improve the image quality in the very dark or very light areas. But I would guess it also depends a lot on the scanner and the operator.
****
Bellow you find your image after working on the original image on bright part of the subject a bit (I hope you do not mind) - I decreased the brightness with 'leves' and added a bit of curves afterwards. The rest of image remained untouched.
It looks indeed differently than your HDR (which looks better to me) but I just wanted to demonstrate the approach.
Al Kaplan
Veteran
The linear response of digital capture seems to make it a problem to keep decent tonal seperation and contrast in both the highlight and shadow areas.
adrianzg
Established
i maximised the dynamic range by putting the film(xp2 in this case) on my diy light table and taking a shot of it using my 5d + 100mm macro. i then processed it in photoshop with recovery set to 100. both shots processed using the same way. i personally like the scanned photo much better cause of the sharpness and contrast.
scanned by the store:
shot w 5d:
scanned by the store:
shot w 5d:
whickus
Established
the potential end result for both techiniques are different, but in this situation, the same result could have been achieved by using curves. when i scanned film (before i started printing and scanning the prints), i always scanned as flat an image as possible, then adjusted the curves, sometimes using several layers - a few for the shadows and a few for the highlights. this made sure that all the details were retained in the scan so that i could work with them later. rule number 1 when scanning is to never trust your scanner. it's a machine that follows the same set of instructions every single time with no consideration given to how you think the photo should look. make sure you like the way it looks (or at least a good working file) before you hit the scan button.
also, darkroom printing could easily achieve these same results. i find selective burning and dodging to be a lot easier and more natural looking than hdr or tone mapping.