Page 1 of 2 12 LastLast
Results 1 to 20 of 25

Thread: Test Image with Uniform Histogram

  1. #1
    dje's Avatar
    Join Date
    May 2011
    Location
    Brisbane Australia
    Posts
    4,183
    Real Name
    Dave Ellis

    Test Image with Uniform Histogram

    Hello all

    I was wondering whether anyone could point me to a test image which has a unifromly distributed histogram ie essentially a horizontal straight line. An 8 bit jpeg would be OK but a 16 bit tif would be better.

    The reason I am looking for this is to try and get an image that I can use to compare the effects on the histogram of different adjustments in PS. eg Shadows/Highlights vs Curves.

    Or am I barking up the wrong tree ?

    Dave

  2. #2
    Momo's Avatar
    Join Date
    Mar 2012
    Location
    Seattle, WA
    Posts
    177
    Real Name
    Darren

    Re: Test Image with Uniform Histogram

    By "image" do you mean photo? Because, that's not going to happen. I have never seen a histogram as a horizontal straight line. If you create a test image and fill it with any solid color or shade of grey the effect will be a spike from top to bottom in the histogram. A gradient will yield a nice curve, but not a straight line. I'm curious to see what other people have to say about a straight line histogram, as I think it's impossible.

  3. #3
    FrankMi's Avatar
    Join Date
    May 2011
    Location
    Fort Mill, South Carolina, USA
    Posts
    6,294
    Real Name
    Frank Miller

    Re: Test Image with Uniform Histogram

    Hi Dave, you might be able to create a B&W gradient that goes from black to white but it would likely be very difficult as the distribution of the pixels changes as you move from black to white. I'm not sure that even if you had such an image that it could tell you anything meaningful when it come to the rendition of a real image as the histogram for any prise winning image won't necessarily be anywhere near close to the histogram of any other.

    A better understanding of how to use histograms to your benefit might be a much more useful exercise.

    Take a look at http://www.cambridgeincolour.com/tut...istograms1.htm and http://www.cambridgeincolour.com/tut...istograms2.htm.

  4. #4
    Moderator Donald's Avatar
    Join Date
    Feb 2009
    Location
    Glenfarg, Scotland
    Posts
    19,913
    Real Name
    Just add 'MacKenzie'

    Re: Test Image with Uniform Histogram

    I had just read the post and was reaching the exact same view as Darren, when I saw his post appearing.

    I suppose it's theoretically possible - What springs to mind are those greyscale charts that you can use to check your monitor (there's one near the bottom of this CiC Tutorial), which go from black at one end to white at the other.

    What you need for a straight line is an image that has everything in it from black to white in exactly equal proportions. Cam't see that happening.

    I'm not sure that I understand what exactly you're trying to do, Dave, Can you not take any image and view the effect of adjustments on the histogram?

    EDIT - Just noticed that Frank has posted as I was preparing the above and his views are along the same lines.

  5. #5
    arith's Avatar
    Join Date
    Sep 2009
    Location
    Burton on Trent, UK
    Posts
    4,789
    Real Name
    Steve

    Re: Test Image with Uniform Histogram

    This is the closest I can get without actually thinking; but then I can't think too hard anymore.

    Test Image with Uniform Histogram

  6. #6

    Join Date
    Dec 2010
    Location
    Greytown, New Zealand
    Posts
    190
    Real Name
    Tim

    Re: Test Image with Uniform Histogram

    Quote Originally Posted by dje View Post
    Hello all

    I was wondering whether anyone could point me to a test image which has a unifromly distributed histogram ie essentially a horizontal straight line. An 8 bit jpeg would be OK but a 16 bit tif would be better.

    The reason I am looking for this is to try and get an image that I can use to compare the effects on the histogram of different adjustments in PS. eg Shadows/Highlights vs Curves.

    Or am I barking up the wrong tree ?

    Dave
    Dave
    There's an excellent image that I've only seen in the context of the Luminous Landscape's video tutorial on Lightroom 4 - it's a download with those videos - and it is used for precisely the purpose you're talking about. I found their explanation of what it was telling me, and my own subsequent tweaking very helpful in understanding exactly how tone curve adjustments work (as opposed to the difference they seem to make) in an image.
    The image they use has both a continuous linear b-w gradient and also a set of 20? patches running from black to white.
    Tim

  7. #7
    dje's Avatar
    Join Date
    May 2011
    Location
    Brisbane Australia
    Posts
    4,183
    Real Name
    Dave Ellis

    Re: Test Image with Uniform Histogram

    Thanks everone for your interest.

    Darren I wasn't expecting a photo like that - it would have to be electronically generated.

    Steve that's about the closest I can get at this stage

    Tim you seem to see what I'm aiming for - unfortunately the Lightroom tutorial videos have to be purchased and as I'm not a Lightroom user this would be a waste.

    Frank and Donald - I probably didn't explain what I was trying to do very well. I'll try and do this with an example (of course my aim could still prove fruitless !!)

    Image 1 : Test Image
    Image 2 : Shadows Adjustment (23%, 25% tonal width)
    Image 3 : Change to histogram with Shadows adjustment
    Image 4 : Curves adjustment
    Image 5 : Change to histogram with Curves adjustment.

    I am trying to use a test image that has a simple and well defined shape when assessing the effects of different adjustments. In this case, there isn't really much difference in the two histogram changes but that won't always be the case.

    Test Image with Uniform Histogram

    Test Image with Uniform Histogram

    Test Image with Uniform Histogram

    Test Image with Uniform Histogram

    Test Image with Uniform Histogram

  8. #8

    Join Date
    Dec 2008
    Location
    New Zealand
    Posts
    17,662
    Real Name
    Have a guess :)

    Re: Test Image with Uniform Histogram

    Quote Originally Posted by dje View Post
    The reason I am looking for this is to try and get an image that I can use to compare the effects on the histogram of different adjustments in PS. eg Shadows/Highlights vs Curves.

    Or am I barking up the wrong tree ?
    Hi Dave,

    Personally, I think it would be more useful to vary these things and look at the changes on a real-world IMAGE, not a histogram.

  9. #9
    dje's Avatar
    Join Date
    May 2011
    Location
    Brisbane Australia
    Posts
    4,183
    Real Name
    Dave Ellis

    Re: Test Image with Uniform Histogram

    Quote Originally Posted by Colin Southern View Post
    Hi Dave,

    Personally, I think it would be more useful to vary these things and look at the changes on a real-world IMAGE, not a histogram.
    Yes Colin, in the end that is what matters and I have of course done this. I guess what prompted my interest was that when doing this on a real image, eg with Shadows vs Curves, I found one gave more pleasing results than the other (and one was easier to adjust than the other) - and being me, I wanted to know why !!

    Dave

  10. #10

    Join Date
    Dec 2008
    Location
    New Zealand
    Posts
    17,662
    Real Name
    Have a guess :)
    Quote Originally Posted by dje View Post
    Yes Colin, in the end that is what matters and I have of course done this. I guess what prompted my interest was that when doing this on a real image, eg with Shadows vs Curves, I found one gave more pleasing results than the other (and one was easier to adjust than the other) - and being me, I wanted to know why !!

    Dave
    Hi Dave,

    Just be aware that for example shadows -v- curves can (and do) overlap - they just offer different approaches to manipulating different tonal ranges.

    Normally I don't use curves - preferring the brightness slider in ACR - but every so often the brightness control just won't push things where I want them, so it's curves to the rescue.

  11. #11
    arith's Avatar
    Join Date
    Sep 2009
    Location
    Burton on Trent, UK
    Posts
    4,789
    Real Name
    Steve

    Re: Test Image with Uniform Histogram

    I thought maybe it could be done in colour; definitely you can get a picture in it by choosing the above gradient and using a special pen that writes x and 255-x luminosity.

    Test Image with Uniform Histogram

    don't know what for though.
    Last edited by arith; 24th March 2012 at 10:44 AM.

  12. #12
    John C's Avatar
    Join Date
    Mar 2009
    Location
    Atlanta, Georgia
    Posts
    157

    Re: Test Image with Uniform Histogram

    I get a linear histogram for this file.

    Test Image with Uniform Histogram

    The minor "unflatness" of the histogram is due to creating a jpg. The .tiff is perfectly flat. If you want a fuzzier looking histogram, maybe you can add some noise.
    Last edited by John C; 24th March 2012 at 11:00 AM.

  13. #13

    Join Date
    Dec 2008
    Location
    New Zealand
    Posts
    17,662
    Real Name
    Have a guess :)

    Re: Test Image with Uniform Histogram

    Quote Originally Posted by John C View Post
    I get a linear histogram for this file.

    Test Image with Uniform Histogram
    Hi John,

    I don't - it's got spikes and holes all through it on my screen.

  14. #14

    Join Date
    Dec 2008
    Location
    New Zealand
    Posts
    17,662
    Real Name
    Have a guess :)

    Re: Test Image with Uniform Histogram

    This is as close as I can get to it -- keeping in mind though that the "varience" might be a limitation of the DISPLAY of the histogram - not the histogram itself.

    Test Image with Uniform Histogram

  15. #15
    John C's Avatar
    Join Date
    Mar 2009
    Location
    Atlanta, Georgia
    Posts
    157

    Re: Test Image with Uniform Histogram

    Quote Originally Posted by Colin Southern View Post
    Hi John,

    I don't - it's got spikes and holes all through it on my screen.
    I think the spikes and holes are just due to autoscaling the vertical axis on whatever program you are using to view the file. What you are seeing is just the noise from the jpeg compression. The .tiff is flat. This was created by filling a blank file with a linear gradient to create an 8-bit file. I created three and composited them as R,G,B to get a 24-bit color file. The original shows a perfectly flat histogram.

    Another problem seems to be the fact that the 24-bit file I uploaded was converted to an 8-bit B/W file somewhere along the way. Not sure how to solve that problem.

    Overall, I can see where a file like this might be helpful in understanding what a editing operation is doing to a photo.

  16. #16
    John C's Avatar
    Join Date
    Mar 2009
    Location
    Atlanta, Georgia
    Posts
    157

    Re: Test Image with Uniform Histogram

    Test Image with Uniform Histogram

    I added some Gaussian noise this time. It helps a bit with the histogram scaling but its not quite as flat.

  17. #17
    dje's Avatar
    Join Date
    May 2011
    Location
    Brisbane Australia
    Posts
    4,183
    Real Name
    Dave Ellis

    Re: Test Image with Uniform Histogram

    Quote Originally Posted by John C View Post
    I think the spikes and holes are just due to autoscaling the vertical axis on whatever program you are using to view the file. What you are seeing is just the noise from the jpeg compression. The .tiff is flat. This was created by filling a blank file with a linear gradient to create an 8-bit file. I created three and composited them as R,G,B to get a 24-bit color file. The original shows a perfectly flat histogram.

    Another problem seems to be the fact that the 24-bit file I uploaded was converted to an 8-bit B/W file somewhere along the way. Not sure how to solve that problem.

    Overall, I can see where a file like this might be helpful in understanding what a editing operation is doing to a photo.
    Thanks for your interest John. Could you elaborate a little more on your method of creating the tif please. in particular - "composited them as R G and B to get a 24 bit colour file"

    Thanks Dave

  18. #18
    John C's Avatar
    Join Date
    Mar 2009
    Location
    Atlanta, Georgia
    Posts
    157

    Re: Test Image with Uniform Histogram

    I used Picture Window Pro but this will probably work in other programs too. 1. I opened a new blank file 1024 by 512 (no reason other than it seemed like a good size). 2. I used the Gradient function using a linear gradient and one cycle (default settings for Gradient function). This filled the blank file with a black to white gradient pretty much as you see it in the file I posted. However, this was only an 8-bit file. So, I created 2 more files just like the first. 3. I used the Combine Channels function to combine the three 8-bit files to create one color (r, g, b) file. Now I had a 24-bit file that was comprised of the three individual 8-bit files (I could probably have used one file for all three channels). 4. The spectrum (histogram) of this file looks a little too pure, so I used the Add Noise function. I added only about 5% noise - just at the verge of visibility. You can play with the amount to suite your taste and check it by looking at the histogram. That's about it.

    Sorry not to get back sooner.

  19. #19
    dje's Avatar
    Join Date
    May 2011
    Location
    Brisbane Australia
    Posts
    4,183
    Real Name
    Dave Ellis

    Re: Test Image with Uniform Histogram

    Thanks John, I'll give it a go if I can find the equivalent proocedures in PS.

    Dave

  20. #20
    John C's Avatar
    Join Date
    Mar 2009
    Location
    Atlanta, Georgia
    Posts
    157

    Re: Test Image with Uniform Histogram

    I think the difficulty might be making sure you have a perfectly linear gradient. If pure white or pure black ends of the gradient are more than a pixel wide, the histogram might show up as having 'tails' rather than being flat. I'm fairly sure that Photoshop has a function that lets you split a color file into r, g, b channels. You might have to replace the individual r, g, b channels with linear gradients and then re-combine the channels. Or something like that. Good luck.

Page 1 of 2 12 LastLast

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •