How to avoid color bleeding ?

[Rudi]

Finally got my DP2M and realized a problem with white letters on a red-magenta background, when doing PP in PSE after RAW conversion in SPP.

Full Image is here : 46MPix / 8Bit JPEG


why is there color bleeding in the text underneath the number 4 ?


It doesn´t show in SPP, but when working on the file in PSE the letters become light magenta.

Anyone an idea how to avoid this effect ?

Any suggestions appreciated.

Thak you in advance,

Rudi.

 

[maceoQ]

This image look very strange and not like a DP2m at all. It's too blue, blurry and seems to have noise reduction applied.
Maybe you can try to export the file as 16Bit TIFF without noise reduction in SPP and check the problematic area again?

 

[Rudi]

This image look very strange and not like a DP2m at all. It's too blue, blurry and seems to have noise reduction applied.
Maybe you can try to export the file as 16Bit TIFF without noise reduction in SPP and check the problematic area again?

Thank you for the reply maceoQ,

did you see the whole image under the link i provided ?

It was the 4th image I took after I got the camera brand new in the morning. So as the rain stoped I went out to play around with it.

It handles a little different than the DP1 or my SDs so may be I have to get used to it and learn how to handle the files from it.

Here are the EXIF datas :

---- File ----
FileName : SDIM0004.jpg
---- EXIF ----
Make : SIGMA
Model : SIGMA DP2 Merrill
Orientation : Horizontal (normal)
XResolution : 300
YResolution : 300
ResolutionUnit : inches
Software : SIGMA Photo Pro 5.5.3.0001
ModifyDate : 2015:02:21 21:54:15
YCbCrPositioning : Co-sited
ExposureTime : 1/100
FNumber : 4.0
ExposureProgram : Program AE
ISO : 100
ExifVersion : 0220
DateTimeOriginal : 2015:02:21 16:58:49
CreateDate : 2015:02:21 21:54:15
ComponentsConfiguration : Y, Cb, Cr, -
ExposureCompensation : -0.7
MaxApertureValue : 2.0
MeteringMode : Unknown
Flash : No Flash
FocalLength : 30.0 mm
FlashpixVersion : 0100
ColorSpace : Uncalibrated
ExifImageWidth : 8400
ExifImageHeight : 5600
SensingMethod : One-chip color area
FileSource : Digital Camera
CustomRendered : Custom
ExposureMode : Auto bracket
WhiteBalance : Auto
FocalLengthIn35mmFormat : 45 mm
SceneCaptureType : Standard
ImageUniqueID : 303034313731323449B9E85431444539
---- MakerNotes ----
SerialNumber : 904xxxxxx
DriveMode : SINGLE
ResolutionMode : HI
AFMode : AF-S
FocusSetting : AF
WhiteBalance : Auto
ExposureMode : Program AE
MeteringMode : Unknown (S)
LensFocalRange : 30
ColorSpace : ROMM RGB
ExposureCompensation : +0,0
Contrast : +1,0
Shadow : -1,0
Highlight : -0,3
Saturation : +0,5
Sharpness : +0,0
X3FillLight : +0,2
ColorAdjustment : 11C+8M
AdjustmentMode : Custom Setting Mode
Quality : 12
Firmware : 1.05.1.7093
Software : SIGMA Photo Pro 5.5.3.0001
AutoBracket : 1 of 3
ChrominanceNoiseReduction : +1,000
LuminanceNoiseReduction : +0,750
ColorMode : FOV Classic Blue

Thanks for your suggestion. 16Bit Tiff might be the way to go.

Also I´ll not use the FOV classic blue mode in the future, because it is not what it promisses. It was just a trial to see what it does. Now, anyhow in SPP the letters are white, but I think PSE is causing the coloration. I have to take a closer look at that too.

Thank you once more,

Rudi.

 

[xpatUSA]

This image look very strange and not like a DP2m at all. It's too blue, blurry and seems to have noise reduction applied.
Maybe you can try to export the file as 16Bit TIFF without noise reduction in SPP and check the problematic area again?

Thank you for the reply maceoQ,

did you see the whole image under the link i provided ?

It was the 4th image I took after I got the camera brand new in the morning. So as the rain stoped I went out to play around with it.

It handles a little different than the DP1 or my SDs so may be I have to get used to it and learn how to handle the files from it.

Here are the EXIF datas :

---- File ----
FileName : SDIM0004.jpg
---- EXIF ----
Make : SIGMA
Model : SIGMA DP2 Merrill
Orientation : Horizontal (normal)
XResolution : 300
YResolution : 300
ResolutionUnit : inches
Software : SIGMA Photo Pro 5.5.3.0001
ModifyDate : 2015:02:21 21:54:15
YCbCrPositioning : Co-sited
ExposureTime : 1/100
FNumber : 4.0
ExposureProgram : Program AE
ISO : 100
ExifVersion : 0220
DateTimeOriginal : 2015:02:21 16:58:49
CreateDate : 2015:02:21 21:54:15
ComponentsConfiguration : Y, Cb, Cr, -
ExposureCompensation : -0.7
MaxApertureValue : 2.0
MeteringMode : Unknown
Flash : No Flash
FocalLength : 30.0 mm
FlashpixVersion : 0100
ColorSpace : Uncalibrated
ExifImageWidth : 8400
ExifImageHeight : 5600
SensingMethod : One-chip color area
FileSource : Digital Camera
CustomRendered : Custom
ExposureMode : Auto bracket
WhiteBalance : Auto
FocalLengthIn35mmFormat : 45 mm
SceneCaptureType : Standard
ImageUniqueID : 303034313731323449B9E85431444539
---- MakerNotes ----
SerialNumber : 904xxxxxx
DriveMode : SINGLE
ResolutionMode : HI
AFMode : AF-S
FocusSetting : AF
WhiteBalance : Auto
ExposureMode : Program AE
MeteringMode : Unknown (S)
LensFocalRange : 30
ColorSpace : ROMM RGB
ExposureCompensation : +0,0
Contrast : +1,0
Shadow : -1,0
Highlight : -0,3
Saturation : +0,5
Sharpness : +0,0
X3FillLight : +0,2
ColorAdjustment : 11C+8M
AdjustmentMode : Custom Setting Mode
Quality : 12
Firmware : 1.05.1.7093
Software : SIGMA Photo Pro 5.5.3.0001
AutoBracket : 1 of 3
ChrominanceNoiseReduction : +1,000
LuminanceNoiseReduction : +0,750
ColorMode : FOV Classic Blue

Thanks for your suggestion. 16Bit Tiff might be the way to go.

Also I´ll not use the FOV classic blue mode in the future, because it is not what it promisses. It was just a trial to see what it does. Now, anyhow in SPP the letters are white, but I think PSE is causing the coloration. I have to take a closer look at that too.

Thank you once more,

Rudi.

Several points from me, Rudi, but first your post but at 800% in FastStone Viewer:



Firstly, the small letters are only about 3px thick. But also look on the edge of the 4 above them. I see either smoothing or possibly JPEG artifacts. One jpeg 'square' is 8x8 pixels - plenty of room for it to fill those tiny letters with confused stuff.

Secondly, if you work in one color space in SPP and save in another color space what you're seeing on-screen is not what gets saved in your output file.

You could save from SPP in ProPhoto 16-bit TIFF (time to make coffee). Set PSE to not do color management. Open the TIFF (it will honor ProPhoto 16-bit TIFF for viewing) . . see what you get. There might still be bleeding if there was a little softness (pixel-peeping) in the original capture.

To avoid JPEG artifacts you can save it from PSE as a PNG after editing and converting to 8-bits . . . if you like big files, that is ;-)

If you could post the RAW somewhere, I would be glad to look more closely.

--
A good post is short on words and long on references. (Me, 2015),
Ted

 

[digi2ap]

Hmm - try to export without any SPP colour noise reduction (slider set to the far left) and see if it makes any difference.

 

[D Cox]

I think it's a combination of JPG artefacts and chroma noise reduction. JPG compression can be as bad a Bayer demosaicing, or worse.

Save from SPP as a 16-bit TIF, and when converting to JPG to show here use only the highest quality. Avoid noise reduction as much as possible.

If you go back to your X3F and try that, it will be interesting to see how it looks.

 

[xpatUSA]

I think it's a combination of JPG artefacts and chroma noise reduction.

Thanks for the reminder about SPP 5+ NR. I understand that we get a good dose of NR at zero. And in 3.5 you get NR whether you like it or not.

So, when I'm bothered with noise - not that it ever appears in my perfect shots - I'll head to RawDigger and analyze there

 

[Rudi]

This image look very strange and not like a DP2m at all. It's too blue, blurry and seems to have noise reduction applied.
Maybe you can try to export the file as 16Bit TIFF without noise reduction in SPP and check the problematic area again?

Thank you for the reply maceoQ,

did you see the whole image under the link i provided ?

It was the 4th image I took after I got the camera brand new in the morning. So as the rain stoped I went out to play around with it.

It handles a little different than the DP1 or my SDs so may be I have to get used to it and learn how to handle the files from it.

Here are the EXIF datas :

---- File ----
FileName : SDIM0004.jpg
---- EXIF ----
Make : SIGMA
Model : SIGMA DP2 Merrill
Orientation : Horizontal (normal)
XResolution : 300
YResolution : 300
ResolutionUnit : inches
Software : SIGMA Photo Pro 5.5.3.0001
ModifyDate : 2015:02:21 21:54:15
YCbCrPositioning : Co-sited
ExposureTime : 1/100
FNumber : 4.0
ExposureProgram : Program AE
ISO : 100
ExifVersion : 0220
DateTimeOriginal : 2015:02:21 16:58:49
CreateDate : 2015:02:21 21:54:15
ComponentsConfiguration : Y, Cb, Cr, -
ExposureCompensation : -0.7
MaxApertureValue : 2.0
MeteringMode : Unknown
Flash : No Flash
FocalLength : 30.0 mm
FlashpixVersion : 0100
ColorSpace : Uncalibrated
ExifImageWidth : 8400
ExifImageHeight : 5600
SensingMethod : One-chip color area
FileSource : Digital Camera
CustomRendered : Custom
ExposureMode : Auto bracket
WhiteBalance : Auto
FocalLengthIn35mmFormat : 45 mm
SceneCaptureType : Standard
ImageUniqueID : 303034313731323449B9E85431444539
---- MakerNotes ----
SerialNumber : 904xxxxxx
DriveMode : SINGLE
ResolutionMode : HI
AFMode : AF-S
FocusSetting : AF
WhiteBalance : Auto
ExposureMode : Program AE
MeteringMode : Unknown (S)
LensFocalRange : 30
ColorSpace : ROMM RGB
ExposureCompensation : +0,0
Contrast : +1,0
Shadow : -1,0
Highlight : -0,3
Saturation : +0,5
Sharpness : +0,0
X3FillLight : +0,2
ColorAdjustment : 11C+8M
AdjustmentMode : Custom Setting Mode
Quality : 12
Firmware : 1.05.1.7093
Software : SIGMA Photo Pro 5.5.3.0001
AutoBracket : 1 of 3
ChrominanceNoiseReduction : +1,000
LuminanceNoiseReduction : +0,750
ColorMode : FOV Classic Blue

Thanks for your suggestion. 16Bit Tiff might be the way to go.

Also I´ll not use the FOV classic blue mode in the future, because it is not what it promisses. It was just a trial to see what it does. Now, anyhow in SPP the letters are white, but I think PSE is causing the coloration. I have to take a closer look at that too.

Thank you once more,

Rudi.

Several points from me, Rudi, but first your post but at 800% in FastStone Viewer:



Firstly, the small letters are only about 3px thick. But also look on the edge of the 4 above them. I see either smoothing or possibly JPEG artifacts. One jpeg 'square' is 8x8 pixels - plenty of room for it to fill those tiny letters with confused stuff.

Secondly, if you work in one color space in SPP and save in another color space what you're seeing on-screen is not what gets saved in your output file.

You could save from SPP in ProPhoto 16-bit TIFF (time to make coffee). Set PSE to not do color management. Open the TIFF (it will honor ProPhoto 16-bit TIFF for viewing) . . see what you get. There might still be bleeding if there was a little softness (pixel-peeping) in the original capture.

To avoid JPEG artifacts you can save it from PSE as a PNG after editing and converting to 8-bits . . . if you like big files, that is ;-)

If you could post the RAW somewhere, I would be glad to look more closely.

--
A good post is short on words and long on references. (Me, 2015),
Ted

Hello Ted,

thanks for your assistance. Turning off "FOV classic blue" in SPP and RAW converted to 16Bit Tiff, double size (337MB file), from there to PNG and then converting to JPG did it.

That takes time, but it´s fantastic which small details the DP2M is capturing so marvelous in a relatively far distance (about 54,5ft / 18m).


... the Scene ...


crop 100%

Still have to experiment with the DP2M files a lot more to get things right .

Many thanks again,

Rudi.

P.S. the X3F RAW file is 54MB and I don´t have enough space on the domain I´m allowed to use for samples .

 

[xpatUSA]

crop 100%

Still have to experiment with the DP2M files a lot more to get things right .

Many thanks again,

Rudi.

P.S. the X3F RAW file is 54MB and I don´t have enough space on the domain I´m allowed to use for samples .

No problem about the X3F, big files eat my satellite bandwidth daily allowance (350MB) anyway.

I wouldn't have thought of saving double-size. Does it improve the result compared to full-size?

--
A good post is short on words and long on references. (Me, 2015),
Ted

 

[Rudi]

crop 100%

Still have to experiment with the DP2M files a lot more to get things right .

Many thanks again,

Rudi.

P.S. the X3F RAW file is 54MB and I don´t have enough space on the domain I´m allowed to use for samples .

No problem about the X3F, big files eat my satellite bandwidth daily allowance (350MB) anyway.

I wouldn't have thought of saving double-size. Does it improve the result compared to full-size?

--
A good post is short on words and long on references. (Me, 2015),
Ted

well - to convert to double-size images is a two edged sword. On one side you get more pixels out of very small details which helps to avoid color bleeding like in the magenta sign on the left, on the other side details below nyquist rate become more blury as you can see in the part of the amarak restaurant menu on the right hand side of the crop.

It depends on what you personally want to do with your image. If you don´t need real large prints or want to look at your images on a monitor, TV-screen or with a beamer, the original size is more than enough. But for large printer output double-size and a good PP program is mandatory. Best results I get with PhotoNinja where you have a "Defringe"-function which helps against color bleeding or Zoner Photo Studio.

PSE has not much to offer in this respect. It can make things worse than they are in the SPP-converted results, might be different in PS or Lightroom.

For resizing I normally use PhotoZoom Pro 5 with S-spline Max and a little tweaking or finetuning of the sharpness settings (default values are to hard sometimes). It takes time to resize to 3 x 2 meters for 180 DPI prints and compared to that SPP is really fast .

Thanks again that you pointed me to the size of the letters. That was what gave me the clue on how I have to set the rendering output in SPP and it worked pretty good I think.

Cheers,

Rudi.