I think I am missing something, and I hope you can help explain. I have 2 questions regarding the use of NI:
1) I see how profiles are made for specific cameras at various ISO's, settings, etc. These profiles are posted and shared on this site. Can you explain if there can be a profile difference between the same specific model, for all these same cameras? (ie, can, or do ALL Canon D-60's produce the exact same profile?) Or, has it been shown that they are "close enough"? Maybe I am missing the point, or can you help explain why this is not a profiling issue?
2) My second question involves trying to make a custom profile for my camera. As instructed, I took an out of focus photo of the target on my computer screen. I proceeded to get rough analysis of the center. I then sampled (for fine tune) each of the 16 boxes (valid ones only ... without clipping, etc). This produced a profile from which I noted each readings for R-G-B (the equalizer readings). I repeated this same procedure 6 times on the same exact photo, each time clearing and deleting the prior profile session. I discovered that I am unable to reproduce any of my profiles ... they all seem different! Please explain if I am doing something wrong, or why I can not seem to get a reproducible profile each time. Is this important?
Thank you! Bri
Profiles for same camera brand?
I will answer two your questions:
1) Yes, there may be a minor difference in the noise properties of cameras of the same model. However, this difference is not significant and therefore using a profile built using another instance of camera of the same model is a good starting point. If you take a look at http://www.neatimage.com/profiling.html, you will see that this method is not the best but good enough. If you want to get highest accuracy in profiling your camera, you need to build these profiles yourself using the calibration target ( http://www.neatimage.com/testtarget.html ) or other image with flat featureless areas.
2) From your description it follows that you did build the profiles in the correct way. (Why didn't you use the auto fine-tuning?)
How much different were the results? Can you give an example?
Vlad
1) Yes, there may be a minor difference in the noise properties of cameras of the same model. However, this difference is not significant and therefore using a profile built using another instance of camera of the same model is a good starting point. If you take a look at http://www.neatimage.com/profiling.html, you will see that this method is not the best but good enough. If you want to get highest accuracy in profiling your camera, you need to build these profiles yourself using the calibration target ( http://www.neatimage.com/testtarget.html ) or other image with flat featureless areas.
2) From your description it follows that you did build the profiles in the correct way. (Why didn't you use the auto fine-tuning?)
How much different were the results? Can you give an example?
Vlad
Example profiles...
Hi Vlad!
Thank you for answering my 2 questions.
#1: "Yes, there may be a minor difference in the noise properties of cameras of the same model. However, this difference is not significant and therefore using a profile built using another instance of camera of the same model is a good starting point," was the answer I was seeking. I am now curious at what the differences may be!
In regards to question & answer #2, here is an example of the profiles I noted (the following small sample to conserve space...reading GREEN equalizer from left to right)
Trial #1 88 164 109 66 70 28 0 0 -37
Trial #2 107 129 81 81 70 13 0 -34 -37
Trial #3 83 95 89 89 79 26 0 0 -42 etc...
I will try your suggestion using "auto fine tune" tonight, on the same photo, and will report back later.
To review, I will first clear & delete all the prior profiles. Then I will take one "rough sample" on the center of the target. Then I will press "auto fine tune". I will then note the readings on the equalizer. I will try and get at least 6 samples for you.
Thank you. Bri
Thank you for answering my 2 questions.
#1: "Yes, there may be a minor difference in the noise properties of cameras of the same model. However, this difference is not significant and therefore using a profile built using another instance of camera of the same model is a good starting point," was the answer I was seeking. I am now curious at what the differences may be!
In regards to question & answer #2, here is an example of the profiles I noted (the following small sample to conserve space...reading GREEN equalizer from left to right)
Trial #1 88 164 109 66 70 28 0 0 -37
Trial #2 107 129 81 81 70 13 0 -34 -37
Trial #3 83 95 89 89 79 26 0 0 -42 etc...
I will try your suggestion using "auto fine tune" tonight, on the same photo, and will report back later.
To review, I will first clear & delete all the prior profiles. Then I will take one "rough sample" on the center of the target. Then I will press "auto fine tune". I will then note the readings on the equalizer. I will try and get at least 6 samples for you.
Thank you. Bri
Bri,
Thank you for the reply.
The difference in the EQ values may be caused by the difference in the areas you select to build a rough profile. This shouldn't be significant, but still this is a factor that may lead to different EQ values. Please send me a couple of profiles to look at.
Thank you,
Vlad
P.S. Just in case: don't forget to use Profile->New menu item when you start building a new profile. This will clear all the existing data and you will be sure that you start from a clean sheet every time.
Thank you for the reply.
The difference in the EQ values may be caused by the difference in the areas you select to build a rough profile. This shouldn't be significant, but still this is a factor that may lead to different EQ values. Please send me a couple of profiles to look at.
Thank you,
Vlad
P.S. Just in case: don't forget to use Profile->New menu item when you start building a new profile. This will clear all the existing data and you will be sure that you start from a clean sheet every time.
Here's some data
Hi Vlad!
"The difference in the EQ values may be caused by the difference in the areas you select to build a rough profile" ... does seems logical to me. My little experiment found that "auto fine tune" seems to be slightly more consistent. However, even though the "manual fine tune" produced wildly different results, it too also seemed produced a pattern (even though it seems to differ from the "auto fine tune"). Here's the requested data:
Manual Fine Tune:
#1 56 106 33 64 70 18 0 0 -43|106 162 128 128 99 36 0 0 -23| 84 90 86 86 74 74 0 0 -33
#2 77 94 91 58 66 17 0 0 -43|148 148 122 122 90 35 0 0 -19|103 103 80 80 70 70 0 0 -45
#3 0 110 99 99 75 8 0 -34 -46|147 193 156 156 127 87 0 0 -28|91 94 84 84 72 72 0 0 -44
#4 43 82 96 96 52 9 0 -36 -51| 82 117 137 137 88 26 0 0 -24|62 62 72 72 59 59 0 0 -50
#5 52 59 64 44 44 8 0 -42 -53|100 100 84 84 59 26 0 0 -36|71 71 53 53 47 43 0 0 -52
#6 35 75 67 41 41 9 0 -37 -68| 62 103 84 84 55 26 0 0 -43|52 57 55 55 44 43 0 0 -55
Auto Fine Tune:
#1 37 30 47 34 34 17 1 -14 -41|46 55 55 54 41 23 4 -9 -47|37 36 42 35 48 26 8 -31 -51
#2 34 28 45 32 33 16 0 -15 -42|45 53 53 53 39 22 3 -9 -47|36 34 41 33 47 25 3 -31 -53
#3 36 29 47 34 33 18 1 -14 -42|46 54 54 54 40 22 4 -9 -47|37 35 42 34 47 26 7 -31 -51
#4 22 17 33 21 23 6 -9 -21 -48|32 39 39 38 27 15 -6 -18 -49|24 23 29 22 34 14 -6 -37 -57
#5 41 34 53 39 39 23 5 -11 -40|52 61 61 56 47 28 8 -6 -45|55 43 50 42 55 32 14 -27 -48
#6 49 43 60 48 48 31 21 -11 -36|57 70 67 60 52 37 11 -2 -38|61 48 59 48 62 30 19 -23 -44
In your experience, which one would you use?
Why did you select it? and how did you know it was the more accurate one?
Would averaging the values help?
Can you explain to me how or why the auto differs from the manual method? or if there was something I did incorrectly?
Sorry this was so long! I'm sure I have more questions. Thank you!
Bri
"The difference in the EQ values may be caused by the difference in the areas you select to build a rough profile" ... does seems logical to me. My little experiment found that "auto fine tune" seems to be slightly more consistent. However, even though the "manual fine tune" produced wildly different results, it too also seemed produced a pattern (even though it seems to differ from the "auto fine tune"). Here's the requested data:
Manual Fine Tune:
#1 56 106 33 64 70 18 0 0 -43|106 162 128 128 99 36 0 0 -23| 84 90 86 86 74 74 0 0 -33
#2 77 94 91 58 66 17 0 0 -43|148 148 122 122 90 35 0 0 -19|103 103 80 80 70 70 0 0 -45
#3 0 110 99 99 75 8 0 -34 -46|147 193 156 156 127 87 0 0 -28|91 94 84 84 72 72 0 0 -44
#4 43 82 96 96 52 9 0 -36 -51| 82 117 137 137 88 26 0 0 -24|62 62 72 72 59 59 0 0 -50
#5 52 59 64 44 44 8 0 -42 -53|100 100 84 84 59 26 0 0 -36|71 71 53 53 47 43 0 0 -52
#6 35 75 67 41 41 9 0 -37 -68| 62 103 84 84 55 26 0 0 -43|52 57 55 55 44 43 0 0 -55
Auto Fine Tune:
#1 37 30 47 34 34 17 1 -14 -41|46 55 55 54 41 23 4 -9 -47|37 36 42 35 48 26 8 -31 -51
#2 34 28 45 32 33 16 0 -15 -42|45 53 53 53 39 22 3 -9 -47|36 34 41 33 47 25 3 -31 -53
#3 36 29 47 34 33 18 1 -14 -42|46 54 54 54 40 22 4 -9 -47|37 35 42 34 47 26 7 -31 -51
#4 22 17 33 21 23 6 -9 -21 -48|32 39 39 38 27 15 -6 -18 -49|24 23 29 22 34 14 -6 -37 -57
#5 41 34 53 39 39 23 5 -11 -40|52 61 61 56 47 28 8 -6 -45|55 43 50 42 55 32 14 -27 -48
#6 49 43 60 48 48 31 21 -11 -36|57 70 67 60 52 37 11 -2 -38|61 48 59 48 62 30 19 -23 -44
In your experience, which one would you use?
Why did you select it? and how did you know it was the more accurate one?
Would averaging the values help?
Can you explain to me how or why the auto differs from the manual method? or if there was something I did incorrectly?
Sorry this was so long! I'm sure I have more questions. Thank you!
Bri
sending...
Hi Vlad!
Thank you for your help. I am very sorry for my misunderstanding.
I redid my trial profiles again, and will now send you the profiles.
Thanks again!
Bri
Thank you for your help. I am very sorry for my misunderstanding.
I redid my trial profiles again, and will now send you the profiles.
Thanks again!
Bri
Bri,
Thank you for the profiles.
In general, I think that there is no problem at all with your results. It is easy to understand why the noise levels vary a bit: you shot in a low ISO mode, where noise levels are generally quite low (which means they are difficult to measure accurately). This, combined with the fact that the image you analyze has a slight gradient (which is not good for profiling), gives some variation in the noise levels. However, if you will open the profile viewer for every profile, you will see that the overall noise level is about the same (about 2.0) - the analysis error is small.
The noise level may be different as compared with the published profiles because your camera mode may be somewhat different. Also, there is always a slight variation in noise properties of particular instances of D60. Anyway, you don't need to match those profiles, you just have to build your own ones.
So, don't worry, what you have seen is most likely not a problem at all. Just try to apply any of the profiles you have (I would select one of those produced with the auto fine-tune function) to your images and see if the result looks good. I believe it will be good.
Vlad
Thank you for the profiles.
In general, I think that there is no problem at all with your results. It is easy to understand why the noise levels vary a bit: you shot in a low ISO mode, where noise levels are generally quite low (which means they are difficult to measure accurately). This, combined with the fact that the image you analyze has a slight gradient (which is not good for profiling), gives some variation in the noise levels. However, if you will open the profile viewer for every profile, you will see that the overall noise level is about the same (about 2.0) - the analysis error is small.
The noise level may be different as compared with the published profiles because your camera mode may be somewhat different. Also, there is always a slight variation in noise properties of particular instances of D60. Anyway, you don't need to match those profiles, you just have to build your own ones.
So, don't worry, what you have seen is most likely not a problem at all. Just try to apply any of the profiles you have (I would select one of those produced with the auto fine-tune function) to your images and see if the result looks good. I believe it will be good.
Vlad
Thank you Vlad!
I will try to work & refine my "auto fine tune" technique. I thought something was incorrect when my profiles differed sharply from the published ones.
Final question, can you explain why my profiles (.dnp files) are 64K (for both auto & manual trials), and the published ones are only 4K? Am I somehow doing too much, or somehow introducing junk to my profiles?
Really appreciate your help. Thank you!
Bri
I will try to work & refine my "auto fine tune" technique. I thought something was incorrect when my profiles differed sharply from the published ones.
Final question, can you explain why my profiles (.dnp files) are 64K (for both auto & manual trials), and the published ones are only 4K? Am I somehow doing too much, or somehow introducing junk to my profiles?
Really appreciate your help. Thank you!
Bri
Bri,
I removed the analyzed image area from the profiles (see the bottom of http://www.neatimage.com/ug/Profiling_options.htm ). You can include or exclude it from the profiles depending on the purpose of this profile.
Vlad
I removed the analyzed image area from the profiles (see the bottom of http://www.neatimage.com/ug/Profiling_options.htm ). You can include or exclude it from the profiles depending on the purpose of this profile.
Vlad