Using ND filters to reduce dynamic range?
- Thread starter ArchRival
- Start date
- Status
left is w/o filter, right is w filter. The histrogram with filter is more shifted towards the middle. Still get a peak at on the left but the peak on the right is gone.
i give up.
GOOD LUCK with your nd filter dynamic range reduction.
maybe you'll win a nobel prize. take it from me though, i've worked with a lot of things to do with cutting down the amount of light intensity through the lens, e.g. ir filter, nd filters.. none of them do what you think they will do.
but hey, maybe it'll float. have a nice day - you can go ahead and be bullheaded. not going to bother.
OF COURSE the peak is shifted towards the middle, one photo is overexposed, one is not. it's as simple as that, not rocket science. but you can go ahead and think that it is. have a nice day.
It's ok. I can see in the first place you don't understand where this is coming from.
Hmm using a filter reduce the intensity of the light. So I guess the key is the exposure time. I believe that if u under expose the histogram would not clip into the dark and white region. Hence achiving the histogram you have that is cluttered in the middle.
Last edited:
yes i can.
it's founded on some boredom in your life, hence you decided that you'd be the great explorer and discover something new in this world.
truth is, if nd filters were great for dynamic range, you bet someone would have already found that out. that much, i know. as for why you're insisting in some numb-skulled way that this will fall through SOMEHOW, by matter of your insistence.. that.. yes, i don't understand where that is coming from, in the face of overwhelming counter-evidence.
it's ok, everyone's entitled to their sandcastles in the sky now and then.
A bright light source will be transmitting more photon per sec then a dim light source and hence using a filter in your case will only reduce the amount of photons reaching your sensor per unit time and therefore came to the conclusion that a nd won't reduce dynamic range but a well use gnd will. I guess this is why nightmare mention you can underexposed. I am assuming that the polymer firm u use is something similar to a nd and not an uneven coated aligned polymer that caused polarization.
Your test is not a controlled test. What you need to do:
Get your hands on a 4 stop ND filter. Borrow one or something.
Find a test subject, best if bright white, against a dark background. Something inanimate. Set up your camera on tripod.
1. Meter the scene. Note the aperture, shutter speed and ISO.
2. Turn to M mode, and apply the settings from #1. Fire a shot.
3. put on the 4 stop ND. Now apply a 4 stop increase in exposure to compensate for the ND. Fire a shot.
Compare the histogram from both pics.
I welcome you to do this test and post screenshots of both histograms here. My money is on the fact they will be almost identical.
To be logical in this theory, lets analyze the behaviour of ND. Neutral Density reduces the light per area into the lens. If you have isolated 3 light source (1lumen, 20 lumen and 100lumen) both will be cut equally by x factor depending on the ND value.
2) Dynamic range is not reduce in theory because all 3 lumens are reduced equally. BUT the first light of 1lumen may disappears bec it is too faint to enter. So what you have achieved making certain area into darkness but the dynamic range is still the same.
3) If you take pic without ND in this case, both 20L and 100L will become a ball of light. Overexpose in their area causing clipping. But the 1L do fine. ND will make 20L fair but 100L smaller ball. (*camera setting is fixed)
4) Now still with ND, if you jack up the shutter time to increase the intake, 20L will brighten and 100L will get worse. 1L may be faint instead of non.
5) In reality, the histogram will change if some lightsource disappeared (as in the 1L) from the sensor, what is bright gets dimmer. However we should not forget that camera meter if not fixed; will average out again.
In conclusion, DR is not reduced but reduces clipping of some bright area. Preventing underexpose overall is not applicable
2) Dynamic range is not reduce in theory because all 3 lumens are reduced equally. BUT the first light of 1lumen may disappears bec it is too faint to enter. So what you have achieved making certain area into darkness but the dynamic range is still the same.
3) If you take pic without ND in this case, both 20L and 100L will become a ball of light. Overexpose in their area causing clipping. But the 1L do fine. ND will make 20L fair but 100L smaller ball. (*camera setting is fixed)
4) Now still with ND, if you jack up the shutter time to increase the intake, 20L will brighten and 100L will get worse. 1L may be faint instead of non.
5) In reality, the histogram will change if some lightsource disappeared (as in the 1L) from the sensor, what is bright gets dimmer. However we should not forget that camera meter if not fixed; will average out again.
In conclusion, DR is not reduced but reduces clipping of some bright area. Preventing underexpose overall is not applicable
wow!
here is magic magic , no need filter to get the peak on the right gone:
iso200, f/2.6, 1/6 second +0.3 ev
iso200, f/2.6 , 1/3 second +1.3 ev
"histogram with 0.3ev is more shifted towards the middle. Still get peaks at on the left but the peak on the right is gone."
this one even better. i think my camera is damn good, the hdr mode:
iso200, f/2.6, 1/13 second -0.7 ev
iso200, f/2.6, 1/3 second +1.3 ev
"histrogram with -0.7ev is EVEN more shifted towards the middle. Still get peaks at on the left but the peak on the right is gone."
histogram checks out your way!!! :heart:
p.s. if don't know how to see that the histogram is due to overexposure, then the nd filter simply made the exposure correct by PREVENTING blown highlights, and shifting the entire thing downwards... then i don't know what to say.
Last edited:
This is where your reasoning went wrong.
In your own terminology, ND filters scale down intensity. But "intensity, hence DR"? Where'd you get that? Intensity ≠ DR, or even proportional to it. DR is simply the range of point intensities across the image or scene.
A properly exposed image may have greater DR than say an overexposed version of the same scene, simply because in the overexposed version, many parts of the image will hit 255 (max on an 8 bit scale). These same parts would be spread out on the histogram of the properly exposed version.
In terms of exposure (ignoring DOF and other effects), adding a ND filter is similar to stopping the aperture down, or increasing shutter speed by the same number of stops as the filter. The histogram shifts just like your examples with the ND filter.
Still doesn't compress the dynamic range though.
Look, you're probably gonna disagree with this as you have with everyone else. That's your prerogative. But if you want to prove your theory, that shot of the lamp is not gonna do it. Try a test chart (with a graduated grey scale) in a controlled lighting setup. With and without ND filter on the same exposure, then changing aperture to get the same effect.
Or shoot a wide DR real scene, with/without ND at the same exposure settings, then by varying aperture/shutter speed. Problem with this is maintaining consistent lighting.
In your own terminology, ND filters scale down intensity. But "intensity, hence DR"? Where'd you get that? Intensity ≠ DR, or even proportional to it. DR is simply the range of point intensities across the image or scene.
A properly exposed image may have greater DR than say an overexposed version of the same scene, simply because in the overexposed version, many parts of the image will hit 255 (max on an 8 bit scale). These same parts would be spread out on the histogram of the properly exposed version.
In terms of exposure (ignoring DOF and other effects), adding a ND filter is similar to stopping the aperture down, or increasing shutter speed by the same number of stops as the filter. The histogram shifts just like your examples with the ND filter.
Still doesn't compress the dynamic range though.
Look, you're probably gonna disagree with this as you have with everyone else. That's your prerogative. But if you want to prove your theory, that shot of the lamp is not gonna do it. Try a test chart (with a graduated grey scale) in a controlled lighting setup. With and without ND filter on the same exposure, then changing aperture to get the same effect.
Or shoot a wide DR real scene, with/without ND at the same exposure settings, then by varying aperture/shutter speed. Problem with this is maintaining consistent lighting.
Last edited:
MORE MAGIC
iso400, f/2.6, 1/25 second.
now with nd filter over the camera lens
iso400, f/2.6, 1/6 second.
WOW!!! the histogram moved. enough to get our rocket scientist here excited.
iso400, f/2.6, 1/25 second. AUTO WHITE BALANCE
iso400, f/2.6, 1/6 second. CUSTOM WHITE BALANCE WITH FILTER ON.
iso400, f/2.6, 1/25 second. CUSTOM WHITE BALANCE WITH FILTER OFF.
oops, looks like did nothing much at all. obviously there will be changes, this is the bottom part of a cheap tianya GND , quite scratched. obviously it will affect the transmittence of light somewhat. i also add the caveat that i did a very anyhow custom wb, can easily see that the custom wb is not spot on for the one with filter on.. but the histogram proves my point.
iso400, f/2.6, 1/25 second.
now with nd filter over the camera lens
iso400, f/2.6, 1/6 second.
WOW!!! the histogram moved. enough to get our rocket scientist here excited.
iso400, f/2.6, 1/25 second. AUTO WHITE BALANCE
iso400, f/2.6, 1/6 second. CUSTOM WHITE BALANCE WITH FILTER ON.
iso400, f/2.6, 1/25 second. CUSTOM WHITE BALANCE WITH FILTER OFF.
oops, looks like did nothing much at all. obviously there will be changes, this is the bottom part of a cheap tianya GND , quite scratched. obviously it will affect the transmittence of light somewhat. i also add the caveat that i did a very anyhow custom wb, can easily see that the custom wb is not spot on for the one with filter on.. but the histogram proves my point.
Last edited:
need to make sure that you state the the custom white balance done on the same object/white card is used. wait our dear friend here says that the curve shift, then it is because of color, but he will refuse to say that.
please see my last two posts. i rest my case. :bsmilie::bsmilie:
in case archrival is going to dispute my settings, here are the files OUT OF CAMERA, without HISTOGRAM, so that he can look at it himself to satisfy his own rocket theory.
#1 original shot, no filter, auto wb
#2 filter, auto wb
#3 another new original shot, auto wb
#4 shot with filter, custom wb done on white wall
#5 shot without filter, custom wb done on white wall
#1 original shot, no filter, auto wb
#2 filter, auto wb
#3 another new original shot, auto wb
#4 shot with filter, custom wb done on white wall
#5 shot without filter, custom wb done on white wall
Thanks for that. Hope to do a test soon, with ND filter or something else equivalent.
To be logical in this theory, lets analyze the behaviour of ND. Neutral Density reduces the light per area into the lens. If you have isolated 3 light source (1lumen, 20 lumen and 100lumen) both will be cut equally by x factor depending on the ND value.
2) Dynamic range is not reduce in theory because all 3 lumens are reduced equally. BUT the first light of 1lumen may disappears bec it is too faint to enter. So what you have achieved making certain area into darkness but the dynamic range is still the same.
3) If you take pic without ND in this case, both 20L and 100L will become a ball of light. Overexpose in their area causing clipping. But the 1L do fine. ND will make 20L fair but 100L smaller ball. (*camera setting is fixed)
4) Now still with ND, if you jack up the shutter time to increase the intake, 20L will brighten and 100L will get worse. 1L may be faint instead of non.
5) In reality, the histogram will change if some lightsource disappeared (as in the 1L) from the sensor, what is bright gets dimmer. However we should not forget that camera meter if not fixed; will average out again.
In conclusion, DR is not reduced but reduces clipping of some bright area. Preventing underexpose overall is not applicable
Okay okay....that's it. Looking for this kind of info. Thanks.
this is called:
counter poison with poison.
sorry, i had to sink to that level. i promise i'll never do it again.
sometimes, when you can't get the message across in a matter of fact way, the next thing to do is to talk on the same level. don't worry, i'm out of here after this last post, proven my point.
Don't be so hard on him lah... Sooner or later he'll be able to figure it out.
BTW, Tiffen makes "contrast reduction" filters but of course they don't really help with capturing highlight and shadow details.
- Status
Similar threads
