Nuking the DNG Camera Profile [Now w/ script]

Joined
Jun 21, 2017
Messages
209
Reaction score
78
Age
51
UPDATE 11/20:
Added some details about a one-step import + profile-nuke script to new post on page 3 of this thread You can see more details in the Readme - GitHub - darana/P4P__color-profiles: Color profiles for the Phantom 4 Pro drone

Update 10/3:

See Dingoz' post page 3 and his dedicated thread Phantom 4 Pro/Adv Lightroom Lens Profile for updated Lens Profiles to work with the raw files that have had the DJI embedded profile removed

UPDATE 9/22:
See page 2 for further work, and I started a repo on Github if you want to download the color profiles I've created as well as the batch files for stripping the profile. GitHub - darana/P4P__color-profiles: Color profiles for the Phantom 4 Pro drone

In light of keeping threads on topic, I'm starting a new thread to discuss stripping out the built in DNG profile settings in the DJI DNG raw files. This is forked from the 20mp not 14mp thread - if you want to discuss if 20MP is false advertising or not, that's the right place to do so. :)

tl;dr I have had some initial success in stripping out the data! Skip down a couple paragraphs for details. Warning, they are nitty gritty and as of now there is no "download this script and it'll magically work" :)

My earlier post in the other thread said:
I'd like to access the truly raw DNG then apply the color profile I've created with the Color Checker and the Lens Profile someone on the board here has generated and see how it compares. In theory with those two profiles we should get a raw file with any lens and barrel distortion corrected for as well as appropriate color response in the demosaicing, but without some of the other preprocessing.

I tinkered for a few minutes with trying to trick Adobe apps into opening the file with no embedded profile. No luck trying to reprocess it with DNG Creator or saving with a multitude of old formats.

A DNG has the EXIF/XMP data embedded in it, so if we access the raw XMP portion of the DNG that should contain the embedded "profile".​

We know that this can be done because other programs that demosaic the DNG files just ignore the settings that Adobe Camera Raw/Lightroom use to apply an initial profile to the file. The problems with this profile seem to be:
  1. Sharpness in the corners is lost, based on reports it seems to be lost at a greater amount than is present in the original raw file.
  2. A basic set of tone curves and color settings seems to be applied, adding in more contrast than is present in the original raw file and possibly resulting in color shifting
While there are options such as using other raw processing tools, for those of us with an existing Adobe workflow that's not a great option. So, next question is, can we really kill these settings? The answer is Yes. And it isn't actually that difficult to use. Or won't be when after a little bit more testing and a quick batch file or two. :)

Is it worth it?
My initial results so far are that re: corner sharpness the difference is VERY noticeable. The difference in the corner sharpness alone is probably going to be worthwhile in terms of adding this to my workflow. Here's a quick screen grab - on the left is the actual corner of the original file with the embedded profile, on the right is the same part of the image framed roughly the same. Both images are zoomed to 100%, so we're not even talking about many hundreds of % of zoom to see a difference.
KERMIT-0025.jpg


In addition, there are definitely contrast and color shifts. Using LAB color mode and the Reference image function in Lightroom there are shifts of between 0.1 and >6(!) across the LAB values over different swatches of a color card.

KERMIT-0026.jpg


Obviously the image then becomes a bit flat and colors may not be quite right, but if we then match it with an appropriate color profile we get a nice appropriate color response. I generated profile from the image. The only change between the left (stripped out all DNG profile settings) and the right virtual copy is the application of the generated profile.

KERMIT-0027.jpg


As a test I did try to apply the DJI P4P lens profile that someone on the boards here generated a while ago, but it doesn't do much. I suspect the profile was generated based on a DNG with the original profile still embedded. There are some very slight adjustments but running through the profile generation process with a non-embedded-profile image should be the next step I think.

How?
Okay, you read my ramblings or you were smart and just skipped down here. :)

Short answer is that I used ExifTool (ExifTool by Phil Harvey) to strip out the offending tags.

Medium answer is that I dumped all of the group0 and group1 tags out of the file with EXIFTool and reviewed them for the likely culprits. The attached file "alltags.txt" has the results of all of the tag names. I nuked these tags to get the current result:
Appear related to color/curves/calibration
-CalibrationIlluminant1= -CalibrationIlluminant2= -ProfileHueSatMapDims= -ProfileHueSatMapData1= -ProfileHueSatMapData2= -ProfileEmbedPolicy= -NoiseProfile=​

Appear related to crop/correction
-DefaultScale= -DefaultCropOrigin= -DefaultCropSize= -BayerGreenSplit= -AntiAliasStrength= -BestQualityScale= -ActiveArea= -OpcodeList1= -OpcodeList3= -DefaultUserCrop= -YCbCrCoefficients= -YCbCrSubSampling= -YCbCrPositioning= -ReferenceBlackWhite=​

See attached file "strippedtags.txt" for a listing as well as the ExifTool command

Long answer? Pending. :)

Current status is that I pretty aggressively nuked those two batches of tags. There was a third group that looked promising but LR couldn't read the files after that. So further testing is required to better isolate the tags that as a group contain the embedded settings.

Current status is also I need to get back to REAL work for now ... but more to come. Wanted to share what I figured out so far in case others want to tinker, too. :)
 

Attachments

  • alltags.txt
    3.8 KB · Views: 536
  • strippedtags.txt
    1 KB · Views: 496
Last edited:
Joined
Jul 26, 2016
Messages
1,539
Reaction score
480
Interesting.

I used the ColorChecker before but found it too small to get a good lock on for colors due to the wide lens. I found some DIY ColorChecker info on a printer forum and made a large one which works much better for me. I can hover over it and snap it as a DNG and then drop it into the ColorChecker Passport DNG Profile Maker software that came with the plastic CC gizmo, and then into Lightroom.

Here: DIY Colorchecker Chart.

I think the colors in the CC are a smaller SRGB color space and many inkjet printers might be able to duplicate them. Make small 4 inch prints on 8x11.5 inkjet paper and glue them onto a black foam core board. That or make a file and take it to a photo printer for a large-poster sized print.
 
Joined
Jan 20, 2017
Messages
603
Reaction score
242
Age
63
Location
UK
the trouble with applying any profile at all is that you are going to get an end result which is a compromise. Barrel distortion/chromatic aberration from the lens isn't a constant and depends on the angle of the photo being taken as well as the lighting etc. DJI's blunt force solution just doesn't work satisfactorily as I've pointed out in other threads - the only 'perfect solution' if you are sending images to print or some other critical application is to treat each DNG individually and develop them as such using an editor which ignores the embedded profile.
 
Joined
Jul 26, 2016
Messages
1,539
Reaction score
480
I use DxO Optics Pro to repair a lot of the drone lens issues and it works well to get it all into alignment, but it is manual as they don't have a profile for the DJI Inpsire series yet. I have to use some DNG to TIFF and then dump it into DxO Optics Pro for distortion correction which it handles well. DxO doesn't do any ColorChecker input either as it seems mostly lens and noise editing and no color work.
 
Joined
Jan 20, 2017
Messages
603
Reaction score
242
Age
63
Location
UK
I use DxO Optics Pro to repair a lot of the drone lens issues and it works well to get it all into alignment, but it is manual as they don't have a profile for the DJI Inpsire series yet. I have to use some DNG to TIFF and then dump it into DxO Optics Pro for distortion correction which it handles well. DxO doesn't do any ColorChecker input either as it seems mostly lens and noise editing and no color work.

correcting errors is one thing but take a look at a P4Pro raw file that hasn't been treated top the embedded profile and you get a bit of a surprise :)
RT.jpg
 
Joined
Jun 21, 2017
Messages
209
Reaction score
78
Age
51
the trouble with applying any profile at all is that you are going to get an end result which is a compromise. Barrel distortion/chromatic aberration from the lens isn't a constant and depends on the angle of the photo being taken as well as the lighting etc. DJI's blunt force solution just doesn't work satisfactorily as I've pointed out in other threads - the only 'perfect solution' if you are sending images to print or some other critical application is to treat each DNG individually and develop them as such using an editor which ignores the embedded profile.

Andy, agreed. I'd far prefer a lens with no distortion. :) Fow now I'm just trying to get the best out of what we have.

If I'm misreading your response, my apologies!

Just to be clear - my goal for now, and what I succeeded with so far, was to "ignore" the embedded profile when processing the image in Adobe Lightroom/Adobe Camera Raw. I didn't post a full shot but after stripping out the EXIF settings the version looked like your example - a noticeably wider FOV with pretty large vignettes in the corners along with shifts in the highlight/shadow curves and color curves.

When we talk about an embedded profile, all that we're really talking about is a number of settings that are fully defined with text values within the EXIF data of the DNG file. So if we "simply" remove the proper EXIF values from the DNG then we will get a raw file that should appear the same as if we open it in another editor such as DXO or Rawtherapee as you mentioned in the other thread.

Again, sorry if I'm misreading your response or for lack of clarity in my initial post above. Processing through a whole bunch at once while tinkering (and procrastinating my actual work :D).
 
Joined
Jan 20, 2017
Messages
603
Reaction score
242
Age
63
Location
UK
Andy, agreed. I'd far prefer a lens with no distortion. :) Fow now I'm just trying to get the best out of what we have.

If I'm misreading your response, my apologies!

Just to be clear - my goal for now, and what I succeeded with so far, was to "ignore" the embedded profile when processing the image in Adobe Lightroom/Adobe Camera Raw. I didn't post a full shot but after stripping out the EXIF settings the version looked like your example - a noticeably wider FOV with pretty large vignettes in the corners along with shifts in the highlight/shadow curves and color curves.

When we talk about an embedded profile, all that we're really talking about is a number of settings that are fully defined with text values within the EXIF data of the DNG file. So if we "simply" remove the proper EXIF values from the DNG then we will get a raw file that should appear the same as if we open it in another editor such as DXO or Rawtherapee as you mentioned in the other thread.

Again, sorry if I'm misreading your response or for lack of clarity in my initial post above. Processing through a whole bunch at once while tinkering (and procrastinating my actual work :D).

no, you are not misreading my response but having to override correct something that shouldn't be there is just adding to the workflow - the simple solution (until we can get the profile 'switchable') is just to use another raw editor :)
 
Joined
Jun 21, 2017
Messages
209
Reaction score
78
Age
51
Re: Profiles -

DXO Optics Pro lens alignment stuff should work great on these files, I'd think? I don't actually know as I haven't used it before. :) In theory we should be able to do quite a bit by generating a new lens profile using Adobe's lens profile generator but starting with a DNG file with the profile stripped out.

Similarly with the color profile. I had generated a color profile previously but it was done with a DNG that had the DJI profile active. It DID improve the color response, it has been part of my default workflow. But it was a very minor change. The screenshots with the color cards above, that profile was generated from the DNG file after I stripped out the embedded profile, and the result was a much better change. That was actually a test card I shot with the Freewell Circular Polarizer, so that's a perfect example of why we want to get away from the embedded profile. The CPL is going to introduce some shift that is different than the stock profile expects.


Interesting.
I used the ColorChecker before but found it too small to get a good lock on for colors due to the wide lens. I found some DIY ColorChecker info on a printer forum and made a large one which works much better for me. I can hover over it and snap it as a DNG and then drop it into the ColorChecker Passport DNG Profile Maker software that came with the plastic CC gizmo, and then into Lightroom.

Here: DIY Colorchecker Chart.

I think the colors in the CC are a smaller SRGB color space and many inkjet printers might be able to duplicate them. Make small 4 inch prints on 8x11.5 inkjet paper and glue them onto a black foam core board. That or make a file and take it to a photo printer for a large-poster sized print.

Gmack, very cool! Without going through to click that link, though, the issue with printing our own color chart is that unless we have a higher end photo printer that has ICC profiles and the matching paper (or a colormunki or x1 pro that can generate profiles for printed media) we'll have about as much control over the actual color output of our printer as we do an uncalibrated monitor. Since you have a colorchecker already I'm guessing you knew that (and, okay, I clicked the link after all and see that the author appears to be using an inkjet w/ factory icc profiles so that's a good start!). But I'd guess that 95% of the folks on the board here don't have access to calibrated printers.

re; The Color Checker Passport being small, yeah, it is. My pics from above were a bit ad hoc - I just shot them yesterday in full sun since I was already outside and was tinkering with the new yaw adjustment setting. I plan on shooting them again soon in a more controlled setup. But what I have found is that being out of focus isn't necessarily an issue with the color checker. In fact when it comes to the grey card side I prefer it out of focus as the smearing helps smooth out for a consistent grey. Net is that if you look at screengrab you'll see the colorchecker is out of focus b/c it is too close, but that means that I was able to get a reasonably large size in the frame.
 
Joined
Jun 21, 2017
Messages
209
Reaction score
78
Age
51
no, you are not misreading my response but having to override correct something that shouldn't be there is just adding to the workflow - the simple solution (until we can get the profile 'switchable') is just to use another raw editor :)

Yeah, I'd far prefer to just be able to say "don't freakin' apply this profile!" But I am not really interested in switching tools right now. I'm deeply embedded with the Adobe ecosystem. I use many other tools when they are better (PTGui, helicon, etc) but Lightroom in particular remains my "home base" and where everything starts and ends.

I'm pretty close to just making a little batch file that will just nuke the profiles. I just want to test the various EXIF lines a bit more to isolate the values which are part of the profile and be sure I'm not 1) missing any and 2) killing any I don't want to kill. At that point a little batch file that can be run after a Lightroom import will take care of it. If I get ambitious maybe I'll even look at adding it in as a script in Lightroom. :)
 
Joined
Jul 26, 2016
Messages
1,539
Reaction score
480
Re: Profiles -

DXO Optics Pro lens alignment stuff should work great on these files, I'd think? I don't actually know as I haven't used it before. :) In theory we should be able to do quite a bit by generating a new lens profile using Adobe's lens profile generator but starting with a DNG file with the profile stripped out.

Similarly with the color profile. I had generated a color profile previously but it was done with a DNG that had the DJI profile active. It DID improve the color response, it has been part of my default workflow. But it was a very minor change. The screenshots with the color cards above, that profile was generated from the DNG file after I stripped out the embedded profile, and the result was a much better change. That was actually a test card I shot with the Freewell Circular Polarizer, so that's a perfect example of why we want to get away from the embedded profile. The CPL is going to introduce some shift that is different than the stock profile expects.



Gmack, very cool! Without going through to click that link, though, the issue with printing our own color chart is that unless we have a higher end photo printer that has ICC profiles and the matching paper (or a colormunki or x1 pro that can generate profiles for printed media) we'll have about as much control over the actual color output of our printer as we do an uncalibrated monitor. Since you have a colorchecker already I'm guessing you knew that (and, okay, I clicked the link after all and see that the author appears to be using an inkjet w/ factory icc profiles so that's a good start!). But I'd guess that 95% of the folks on the board here don't have access to calibrated printers.

re; The Color Checker Passport being small, yeah, it is. My pics from above were a bit ad hoc - I just shot them yesterday in full sun since I was already outside and was tinkering with the new yaw adjustment setting. I plan on shooting them again soon in a more controlled setup. But what I have found is that being out of focus isn't necessarily an issue with the color checker. In fact when it comes to the grey card side I prefer it out of focus as the smearing helps smooth out for a consistent grey. Net is that if you look at screengrab you'll see the colorchecker is out of focus b/c it is too close, but that means that I was able to get a reasonably large size in the frame.

Yes, I use a color calibrated printer, actually 6 of them so not a problem here along with an i1 Photo Pro 2 head and BasICColor's calibration software for it. I made the TIFF file following the link's instructions for the color patches and it came out pretty much a spot-on copy of the ColorChecker chart. I should be able to take that image to any printer and they'd make a similar copy, or
I'd hope!

As to DxO Optics Pro 11, it can do a lot of the correction stuff as well as the vignetting, straightening of the horizon, and fix the perspective of building sides and light poles. Pretty remarkable program, just wish I could go from the Inspire 2 DNG into it without making a TIFF first. Maybe they will make a profile set for the Inspire 2 camera someday. I believe they have the lens database already as they have some micro 4/3 lenses, just need the camera body info and DNG stuff.

Here is Andy K's image above run through it, albeit cropped due to perspective tilt. DxO applied a lot of barrel correction, horizon, sharpening, and edge vertical perspective. Must have been a fisheye maybe for all the barrel it used.

RT_DxO-1.jpg
 
Last edited:
  • Like
Reactions: DaRana
Joined
Jun 21, 2017
Messages
209
Reaction score
78
Age
51
Yes, I use a color calibrated printer, actually 6 of them so not a problem here along with an i1 Photo Pro 2 head and BasICColor's calibration software for it. I made the TIFF file following the link's instructions for the color patches and it came out pretty much a spot-on copy of the ColorChecker chart. I should be able to take that image to any printer and they'd make a similar copy, or
I'd hope!

Yeah you've got me thinking about making one too. :) I'll see if it makes a difference once I reshoot the small card in a more controlled environment.

Good demo on the file Andy_K posted. When I can find the time I'm going to try to generate a lens profile using Adobe's Lens Profile tool. I'll try it on that file, too, we can see how it compares.

Having to dump to TIFF first is a real pain. I typically intermediate to JPG when I'm doing the HDR panos just because I don't want to deal with the extra file sizes and processing time - it already takes hours to build one of the big ones! I know I may end up regretting it some day but there's only so much you can do. :)
 
Joined
Jan 20, 2017
Messages
603
Reaction score
242
Age
63
Location
UK
Here is Andy K's image above run through it, albeit cropped due to perspective tilt. DxO applied a lot of barrel correction, horizon, sharpening, and edge vertical perspective. Must have been a fisheye maybe for all the barrel it used.

View attachment 85993

You are still cropping out lots of the image though - it's a brute force solution that really isn't much better than the ACR offering

You've also wiped out the highlights (front left of curved building)

Here's the same image (very quickly) developed in Rawtherapee with no cropping - yes, the vignetting needs a little more work but at least it is a 20mp image :)
DJI_0773sm.png
 
Joined
Jun 21, 2017
Messages
209
Reaction score
78
Age
51
Interesting.

I used the ColorChecker before but found it too small to get a good lock on for colors due to the wide lens. I found some DIY ColorChecker info on a printer forum and made a large one which works much better for me. I can hover over it and snap it as a DNG and then drop it into the ColorChecker Passport DNG Profile Maker software that came with the plastic CC gizmo, and then into Lightroom.

Here: DIY Colorchecker Chart.

I think the colors in the CC are a smaller SRGB color space and many inkjet printers might be able to duplicate them. Make small 4 inch prints on 8x11.5 inkjet paper and glue them onto a black foam core board. That or make a file and take it to a photo printer for a large-poster sized print.

Hey GMack, so I came full circle on this. :) Thanks for the suggestion. I just made a modified version of this last night. I happened to have a smallish piece (8.5" x 7.5") of 1/2" plywood in my car from a project -- when I went to fly Friday it occurred to me that it is just large enough for the p4p to sit on for calibration/takeoff. It is also a perfect size to then pick up, put on my lap, and be a small little "desk" for the remote control when I'm sitting in my folding chair.

Finished+stained the board and printed out a color checker with the addition of a 8" x 1.5" grey card strip. I laminated it and epoxied that to the board last night. It's glossier than I'd like due to the laminate but should be good enough for a ballpark/better than nothing. I've also been testing turning the gimbal down about 45 degrees onto a card on the ground before takeoff - using either white for ETTR or neutral grey for centerline on the ground - and it seems to give a decent exposure and not get modified much if at all for macro light falloff. Hope is that this can serve for that as well. :)

So yeah, thanks again for sharing the suggestion. :) Overall it isn't massively larger than the regular color checker - if it works well I may make a 2nd version a little bigger.
 
Joined
Jul 26, 2016
Messages
1,539
Reaction score
480
Hey GMack, so I came full circle on this. :) Thanks for the suggestion. I just made a modified version of this last night. I happened to have a smallish piece (8.5" x 7.5") of 1/2" plywood in my car from a project -- when I went to fly Friday it occurred to me that it is just large enough for the p4p to sit on for calibration/takeoff. It is also a perfect size to then pick up, put on my lap, and be a small little "desk" for the remote control when I'm sitting in my folding chair.

Finished+stained the board and printed out a color checker with the addition of a 8" x 1.5" grey card strip. I laminated it and epoxied that to the board last night. It's glossier than I'd like due to the laminate but should be good enough for a ballpark/better than nothing. I've also been testing turning the gimbal down about 45 degrees onto a card on the ground before takeoff - using either white for ETTR or neutral grey for centerline on the ground - and it seems to give a decent exposure and not get modified much if at all for macro light falloff. Hope is that this can serve for that as well. :)

So yeah, thanks again for sharing the suggestion. :) Overall it isn't massively larger than the regular color checker - if it works well I may make a 2nd version a little bigger.

I made one on my 17" pigment printer and it turned out surprisingly good using the Wiki hex numbers per the link. Spray mounted it to foam core board, and gave it a satin clear coat too with print spray. Damp grass isn't playing nice with the foam core (cardboard) though so it needs something better to mount it on. Plexiglass or Lexan sheet maybe.

I saw Costco makes canvas prints. They have a 20x30 inch for $58 (Mounted I think in a stretcher frame?). Might be able to get it cheaper without the frame and just roll it up and use it as a landing mat?

I just started using it in Resolve 14 as it has the calibration for the ColorChecker chart within it. Should make that a lot easier for video too. Just hit the C1 switch to flip the camera down on landing to capture it while in video mode. I can see why the guy said it has to be BIG with these wide angle drone lenses.
 
Joined
Jun 21, 2017
Messages
209
Reaction score
78
Age
51
I made one on my 17" pigment printer and it turned out surprisingly good using the Wiki hex numbers per the link. Spray mounted it to foam core board, and gave it a satin clear coat too with print spray. Damp grass isn't playing nice with the foam core (cardboard) though so it needs something better to mount it on. Plexiglass or Lexan sheet maybe.

I saw Costco makes canvas prints. They have a 20x30 inch for $58 (Mounted I think in a stretcher frame?). Might be able to get it cheaper without the frame and just roll it up and use it as a landing mat?

I just started using it in Resolve 14 as it has the calibration for the ColorChecker chart within it. Should make that a lot easier for video too. Just hit the C1 switch to flip the camera down on landing to capture it while in video mode. I can see why the guy said it has to be BIG with these wide angle drone lenses.

I did the same as you described and as in the link. Made it in Ps with the hex values then printed on my epson 3880. The only dif was I printed in luster paper as I had the glossy ink active and didn't want to waste it in the switch to matte for the first test version. Plus I knew I planned to laminate it so it was gonna gloss anyways. :) It was very close - with the only ones visibly off being the reds in row 2 + 3, and only slightly so.

Yeah, I can see foamcore not surviving well in damp grass. That's part of why I went with the wood, that and I just randomly had it sitting there. I didn't trust the spray mount on the wood + damp hence why I went with epoxy. But I know not everyone has epoxy setup sitting around. It's funny how we use the tools we have and convince ourselves it's best for the job. :)

A piece of acrylic is an interesting alternative. I may try that. At a smaller size at least the acrylic at 1/8" should be as good as the 1/2" birch ply as far as flex goes. After I beta test it today I'm going to toss a coat of marine varnish on the bottom and edges of this piece of ply which should make it pretty much invulnerable for the occasional damp for a long time.

Not sure about then canvas. I'd worry about dealing with a damp piece of canvas. The ink should be fine but it could get moldy etc. Part of making mine he size I did for this first version was so it could fit in my hard case or backpack. I'd wonder if the Costco canvas prints have a wide enough gamut.

My goal is to make it equally functional as a launching pad, color/gray card, and as that little table in my lap. It's a small thing but the little table big enough to put the controller on flat makes a huge difference in manual flight to help smooth out those last little bits for great footAge. I don't mind hand catching, this one is too small to really trust as a landing pad.

I saw that Resolve supports the color checker natively. I'm still pretty neophyte at video editing and color grading so I'm still mostly sticking with premiere - learn one tool really well and the principles behind it there and then expand. But with native colorchecker plus this thing, hmm, might be worth tinkering with it in the color grade portion of workflow.
 
Joined
Mar 3, 2017
Messages
479
Reaction score
179
Age
57
I have a suggestion. All this creation of a huge "color checker" is pointless if it's not color accurate. That's the point of this exercise. The passport color checker is a color accurate source that software can look at and correct color based on known values. That's the color the app can use as a reference. Printing up all these hand made checkers I guarantee is an exercise in futility. Unless your printer (the epson) has gone thru a color calibration process and using generate icc profiles, there is no point. Same goes double for Costco. They are horribly inaccurate ESPECIALLY canvas prints. What looks good for a family pic is subjective. Making "color checkers" is a science that needs accuracy.

Just buy a passport color checker. Hover your drone when you take off. Make sure you are in Raw. And white balance is always manual. Take a pic of your color checker. Then you are off.

Unless you work in a lab and can generate prints with that level of accuracy you are just adding to your grief.

All clever ideas here though...
 
Joined
Jul 26, 2016
Messages
1,539
Reaction score
480
I have a suggestion. All this creation of a huge "color checker" is pointless if it's not color accurate. That's the point of this exercise. The passport color checker is a color accurate source that software can look at and correct color based on known values. That's the color the app can use as a reference. Printing up all these hand made checkers I guarantee is an exercise in futility. Unless your printer (the epson) has gone thru a color calibration process and using generate icc profiles, there is no point. Same goes double for Costco. They are horribly inaccurate ESPECIALLY canvas prints. What looks good for a family pic is subjective. Making "color checkers" is a science that needs accuracy.

Just buy a passport color checker. Hover your drone when you take off. Make sure you are in Raw. And white balance is always manual. Take a pic of your color checker. Then you are off.

Unless you work in a lab and can generate prints with that level of accuracy you are just adding to your grief.

All clever ideas here though...

It would be best if someone used a calibrated printer. I have six printers and a boatload of x-rite calibrators and densitometers so it's not an issue for me. Many do not calibrate anything and just eyeball it in, and maybe on an uncalibrated monitor too. The link I posted to is in a printer's forum so I'd guess they all use them as well. The large chart with the drone in the link is pretty much spot on to my retail x-rite ColorChecker too. One can always try with their printer and see how close it matches, or give it to a lab to do.

Fwiw, the colorspace of the ColorChecker falls well within the smaller sRGB color space and most decent printers can probably print it. I put it into ColorThink Pro and it formed the star pattern of the ColorChecker image colors and it is within the 2D and 3D boundary of the sRGB standard profile. Given that, if Costco uses a calibration service like they did with their chemical printers with film and paper, I'd wager they could come up with a pretty decent match if one gives them a TIFF with the Wiki ColorChecker numbers as an image to print.

The whole point of the exercise is to come up with a standard and a lot larger one too as mentioned in the link. If one were to buy a ColorChecker in a 20x30 inch size, it would cost hundreds if not more.

ColorChecker in ColorThink Pro 3.0.3 with sRGB colorspace surrounding it. Adobe 1998 RGB would be a lot larger and wasted on the CC image plus fewer printer's to ink that colorspace.

CC_2D.jpg CC_3D.jpg
 
  • Like
Reactions: DaRana and ravedog

New Threads

Forum statistics

Threads
140,868
Messages
1,451,350
Members
102,709
Latest member
mbah2017