Suppose you had nothing but noise and applied no NR then the jpg would be maximally large because minimal correlation would exist in the image. I understand what you are asking but not sure a jpg is the best answer.
That's right. In this analysis I don't care whether it is noise or real detail.
Exactly. I can't imagine that checking the size of jpg files is the best/only answer for what I want to do. However, I'm not aware of any other tool to do this.
Don't know about an existing program, but if you are willing to do a small programming effort then one can use a variation of a statistical formulation. We can talk offline about it if you are interested.
Sorry, I don't know enough about programming. Thanks anyway.
Hi,
.
Yes, I think this is exactly what I want. Grayscale should be fine and I have used ImageJ (no plugins) years ago. How can I get your plugin and how is it used/installed? Are there any parameters in the process that can be modified?Hi,
I decided to put the method I was thinking about in an ImageJ plugin. ImageJ is a freely available image processing toolbox that can be downloaded online. I can make the plugin available to you if you are interested and if this is what you want. Need to put in some enhancements - such as right now it is only working for grayscale images; need to add color. This is the first time I ever programmed in ImageJ so I have to learn some stuff here also
I can send you a link to download when it is ready. I think mid next week. You would just have to put it in the plugins directory of ImageJ and that's it. Currently, there are no parameters, but if you have something in mind, please let me know.
OK, thank you.
I was just wondering what exactly your program is measuring and whether e.g. the "radius" for microcontrast measurement could be modified.
I plan to use my 100L macro with a 60D to reproduce flat objects and wanted to know where the sweet spot with this particular setup is. It could obviously be used to analyze the effect of f-ratio on lens performance (overall/center/border) at different focus distance.
Of course this could also be used for a lot of other things, quantification of noise and blur from camera/subject movement included.
Yes, of course.
I think he just wants to quantify the difference in intensity between neighboring pixel intensities. You could do this with a high-pass filter, and calculate the standard deviation, but for an image where most of the frame is expected to be out-of-focus, this can downplay the maximum sharpness. The histogram of the high-pass will give more information, but not a monolithic value.
That I lacked a clue as to what he was driving at...
John
I think that your jpeg size is a great idea. it is equivalent to a low pass filter; as jpeg quality is increased higher frequency components are retained so changes in file size with jpeg quality will show aggregate high frequency components.
Don't think that's possible, because changing the f-stop or ISO (exposure) in itself will affect the amount of detail in the image ('real' detail if DoF changes and 'noise detail' if ISO changes), but it could be useful when selecting the shapest image amongst several images of the same subject with e.g. to low shutterspeed. Think that Nikon's 'best shot selector' works like that, by simply comparing the JPEG file sizes.
What if one wants to compare images of different dimensions (widths and height)? Is file size still a good measure? What if one wants to compare different sized patches within the same image, as Mattr wants to do by changing radius?
Well, selecting the sharpest/lowest noise image is exactly what I want to use the program for. I guess my use of "detail" causes a lot of confusion.
Interesting, I have to look into it.
Is your plugin basically measuring the same thing that would be measured by analyzing jpg file sizes?
Unfortunately, I'm not the one to write such a script. Isn't it surprising that this doesn't seem to exist yet?
