Reading mode: Light Dark

7(i) RAW interpolation

Go to last post
A

Auke Nauta

Active member
Messages
99
Reaction score
4
Location
US
Hi,

I'm looking into alternatives to DIVU, when it comes to interpolating the RAW image files from these Minolta camera's.

BTW, it's a shame Minolta did not reduce the file size for .RAW files.

As the actual AD converter is 12-bit, they could have stored a 4-pixel data in 6 bytes. Yet Minolta chose to store each pixel data in 2 bytes, wasting 33% space and increasing the storage time to the Flash media :(

But, I was talking about RAW image interoplation:

DIVU is using bilinear interpolation which is fast (probably used in-camera as well) yet not very accurate.

Alternatives exist which are more computational costly but give substantially better results.

Does anyone know if such an application already exists?

If not, I'm going to write it myself (which I will probably do anyway...)
Anyone interested?

Greetings,
Auke Nauta
 
Mssrs. Russel Bozian, Paul Hansen & Patrick Polan are the ones who have extensive knowledge in this subject. They mostly reside in Bryan Biggers' forum but increasingly rarely talk about this subject which very few people even begin to understand.

You have all the encouragement from the rest of us here. looking forward to press some buttons in due course.

Gordon Chau
Auke Nauta said:
Hi,

I'm looking into alternatives to DIVU, when it comes to
interpolating the RAW image files from these Minolta camera's.

BTW, it's a shame Minolta did not reduce the file size for .RAW files.
As the actual AD converter is 12-bit, they could have stored a
4-pixel data in 6 bytes. Yet Minolta chose to store each pixel data
in 2 bytes, wasting 33% space and increasing the storage time to
the Flash media :(

But, I was talking about RAW image interoplation:
DIVU is using bilinear interpolation which is fast (probably used
in-camera as well) yet not very accurate.

Alternatives exist which are more computational costly but give
substantially better results.

Does anyone know if such an application already exists?

If not, I'm going to write it myself (which I will probably do
anyway...)
Anyone interested?

Greetings,
Auke Nauta
Click to expand...
 
http://webpages.charter.net/bbiggers/DCExperiments/html/minolta_raw_format.html

By the way, I'm not convinced that DIVU uses linear interpolation.

Jake's and Paul's documentations are fairly complete regarding what's known to the civilized world about MRW. Russel proposed a compression scheme a while ago, but I've to confess that I've spent very few hours working on that stuff in the last months, and that I've not even started working on compression support despite what I promised earlier this year.

-- Patrick
 
Auke Nauta said:
But, I was talking about RAW image interoplation:
DIVU is using bilinear interpolation which is fast (probably used
in-camera as well) yet not very accurate.
Click to expand...
Recommended reading:
http://ise.stanford.edu/class/psych221/99/tingchen/main.htm

Since most of the in-camera processing is probably done by dedicated hardware (some custom LSI microcontroller built around a MIPS core), they could possibly be doing something smarter than linear interpolation.

-- Patrick
 
AMonroy said:
In what does it program it can be?
Click to expand...
Probably most of the MRW converters. Have a look to:
http://www.pbase.com/crowning/linkssoftware

Image shifter is nice to get an idea of the difference that the interpolation method makes: load a raw file "full scale", zoom in into some portion of the image, and then toggle the interpolation method.

DIVU has a fairly good interpolation (and sharpening) engine, but I don't know for sure what algorithms are used by this mysterious program.
AMonroy said:
Does some plug ins exist for Photoshop?
Click to expand...
I don't think so.

-- Patrick
 
Hi Patrick,

I've not still been able to buy the D7i, because it has not arrived to my country. Meanwhile I'm manipulating the pictures of some users of the forum with Divu 2 and Photoshop 7. However still I've not been able to get a picture in format MRW of the D7i.

Do you know where I can get it?

Thank you for your help,

AMonroy
 
Patrick, thank you for your help.

I already have the two files MRW. The picture of the D7i was captured with iso 200 and with flash and it is very clear. However it is ugly and although the picture has many colors.

The picture of the D7 was captured to iso 100 and it is specially clear. When I use the Divu 2 I obtain a more blurred picture that when I use ishifter. However the Divu is a much more elaborated program because it allows me to know the characteristics of the picture (It can read exif 2.2), the balance of white when it was captured,... It also associates a file icc, EXIF and DOPF when exporting the tiff.

It is a pity that the program ishifter fails in these areas.

Do you know other programs with interpolation variable number of gradients?

An intermediate solution can be to use ishifter (I suppose that in a future it supports icc, exif, Dopf...) and to open the picture with Photoshop 7. It is necessary to associate him a file icc. In this forum I have seen that they have gotten a file icc of the D7i that works in Photoshop, but it is only correct for pictures tiff and jpg. If somebody had the icc for pictures raw of the D7i, we would have a great help.

I have proven a little in Photoshop and I believe that a good way to export the files raw is with little contrast and little saturation. More details are exported this way the Photoshop. There then the picture is manipulated and we can get very good results.

What do you say?

Greetings,

AMonroy
 
AMonroy said:
The picture of the D7 was captured to iso 100 and it is specially
clear. When I use the Divu 2 I obtain a more blurred picture that
when I use ishifter. However the Divu is a much more elaborated
program because it allows me to know the characteristics of the
picture (It can read exif 2.2), the balance of white when it was
captured,... It also associates a file icc, EXIF and DOPF when
exporting the tiff.

It is a pity that the program ishifter fails in these areas.

Do you know other programs with interpolation variable number of
gradients?
Click to expand...
Working on it right now!
AMonroy said:
An intermediate solution can be to use ishifter (I suppose that in
a future it supports icc, exif, Dopf...) and to open the picture
with Photoshop 7. It is necessary to associate him a file icc. In
this forum I have seen that they have gotten a file icc of the D7i
that works in Photoshop, but it is only correct for pictures tiff
and jpg. If somebody had the icc for pictures raw of the D7i, we
would have a great help.
Click to expand...
I'm working in the Digital Imaging field, and as such, have access to great profile making equipment.
Soon, when I get my 7i, I'll give it a shot.

Please note that every colour device requires a specialised colour profile, that is no 2 devices are exactly the same.

So when I create a custom profile for my camera, you can use it, but results will be optimal only on my system.
You probably know all of this...

Greetings,
Auke Nauta
 
Patrick, have you made Image Shifter? Because I have just realized when I was reading the text that accompanies to the program.

Of all the programs that I have proven, yours is the one that more I like it. The quality is fantastic and it is very quick.

I have some doubts: if I want to maintain the biggest dynamic range, I suppose that it should put to the minimum the intensity, the saturation and the contrast. But, and the gamma? My intention is to substitute the DIVU 2 for your program. I have usually always used Photoshop. I find the configuration of the color in Photoshop more difficult than in Corel Draw 10. When caring an image processed by Image Shifter to the Corel Draw, when not having an incorporate icc, what icc should I choose? There is several sRGB.

Could it export the whole range of colors and dynamic range of of the D7i in the file tiff of 48 bits with Image Shifter?

Do you know if there is some way to incorporate the information EXIF 2.2 manually to the files tiff that exports Image Shifter?

Is there some way to introduce a numeric value for the temperature of the color?

I have checked with other programs to process files MRW that only process the files of the D7. The files of the D7i process them corrupt.

Why Image Shifter processes the files of the D7 and the D7i appropriately? Is the program able to distinguish when the file is of the D7 or of the D7i? Both have different icc. I suppose that they should be processed in a different way.

Thank you and congratulation for your program,

AMonroy
 
AMonroy said:
Patrick, have you made Image Shifter?
Click to expand...
Yes.
AMonroy said:
if I want to maintain the biggest dynamic
range, I suppose that it should put to the minimum the intensity,
the saturation and the contrast. But, and the gamma?
Click to expand...
There is no general answer to this. Having a look to the histogram may help, but in my program, the histogram window only shows luminance distribution, in arbitrary units.
AMonroy said:
My intention
is to substitute the DIVU 2 for your program.
Click to expand...
Well, DIVU has much more consistent results. I have to stuggle with different parameters with each new image in Image Shifter. However, as a first processing step that's OK.
AMonroy said:
When caring an image
processed by Image Shifter to the Corel Draw, when not having an
incorporate icc, what icc should I choose? There is several sRGB.
Click to expand...
You might find answers here:
http://www.srgb.com/index.html
AMonroy said:
Could it export the whole range of colors and dynamic range of of
the D7i in the file tiff of 48 bits with Image Shifter?
Click to expand...
I hope so. In order to fight noise, try neatimage:
http://www.neatimage.com

(I'm the URL bunny)
AMonroy said:
Do you know if there is some way to incorporate the information
EXIF 2.2 manually to the files tiff that exports Image Shifter?
Click to expand...
I'll try to add that in a future revision. I agree that this would be a nice addition.
AMonroy said:
Is there some way to introduce a numeric value for the temperature
of the color?
Click to expand...
No.
AMonroy said:
I have checked with other programs to process files MRW that only
process the files of the D7. The files of the D7i process them
corrupt.
Click to expand...
They haven't been updated yet. It would be good if people could put a few D7i MRW files on some file server.
AMonroy said:
Why Image Shifter processes the files of the D7 and the D7i
appropriately? Is the program able to distinguish when the file is
of the D7 or of the D7i? Both have different icc. I suppose that
they should be processed in a different way.
Click to expand...
Right! This is very true, and for now Image Shifter is using the D7 ICC profile data. However it produces different colors than DIVU because I'm averaging processed colors with unprocessed ones in the output. Again, this is totally arbitrary.

-- Patrick
 
Auke Nauta said:
Please note that every colour device requires a specialised
colour profile, that is no 2 devices are exactly the same.
So when I create a custom profile for my camera, you can use it,
but results will be optimal only on my system.
You probably know all of this...

Greetings,
Auke Nauta
Click to expand...
it would be so nice of you to tell us how to make our own icc profile.
 
dream said:
Auke Nauta said:
Please note that every colour device requires a specialised
colour profile, that is no 2 devices are exactly the same.
So when I create a custom profile for my camera, you can use it,
but results will be optimal only on my system.
You probably know all of this...

Greetings,
Auke Nauta
Click to expand...
it would be so nice of you to tell us how to make our own icc profile.
Click to expand...
Well, it involves taking a picture (under controlled conditions) of a colour chart. The colours of the chart should be known (La*b* values).
The captured RGB data can then be related to the actual colour values.

Basically that's all there is in a profile, the relationship between device independent Lab and (usually) device dependent RGB.

To generate a profile you need special software. For instance, when profiling a printer, you need to measure the Lab values using an expensive spectrofotometer in order to get the best results...

But, quite reasonable results can also be obtained using your own desktop scanner in combination with colour matching programs like Profile Prism (www.ddisoftware.com).

I can recommend this inexpensive program to everyone. I compared the results with results from much more expensive software and I must say it came quite close.

Also, Profile Prism has an excellent help file (also on their web site) staing clearly how to generate a custom profile.

Greetings,
Auke Nauta
 
Thank you Patrick.

The D7i has not still arrived to my country (Spain), but when I have it I could send you some file MRW.

I hope you warn me when you have a new version of Image Shifter.

A greeting,

AMonroy
 
Hello Patrick,

Just been looking for the best workflow to produce most consistent colors and sharpest images and I came across ishifter. Looks great and I cant wait to get my D7i to start experimenting with ishifter. Something else I found, was software to make your own ICC profiles. From what I understood, a snapshot of a color target is processed by software to produce a profile. That snapshot must be as unprocessed as possible and gamma has to be 1.0. Is it possible to do so with iShifter?

Congratulations for ishifter. It's ammasing to see one person make up for the shortcoming of a multinational.

Peter Marina

PS: As soon as I get my D7i, I could setup an FTP and volunteer some files. Also, here is the links to 3 free ICC profilers. The first one came up more then once, could be a sign accuracy...

http://www.aim-dtp.net/aim/calibration/xlprofiler/index.htm
http://www.littlecms.com/iphoto/download.htm
http://www.littlecms.com/profilers.htm
Patrick Porlan said:
Patrick Porlan said:
Patrick, have you made Image Shifter?
Click to expand...
Yes.
Patrick Porlan said:
if I want to maintain the biggest dynamic
range, I suppose that it should put to the minimum the intensity,
the saturation and the contrast. But, and the gamma?
Click to expand...
There is no general answer to this. Having a look to the histogram
may help, but in my program, the histogram window only shows
luminance distribution, in arbitrary units.
Patrick Porlan said:
My intention
is to substitute the DIVU 2 for your program.
Click to expand...
Well, DIVU has much more consistent results. I have to stuggle with
different parameters with each new image in Image Shifter. However,
as a first processing step that's OK.
Patrick Porlan said:
When caring an image
processed by Image Shifter to the Corel Draw, when not having an
incorporate icc, what icc should I choose? There is several sRGB.
Click to expand...
You might find answers here:
http://www.srgb.com/index.html
Patrick Porlan said:
Could it export the whole range of colors and dynamic range of of
the D7i in the file tiff of 48 bits with Image Shifter?
Click to expand...
I hope so. In order to fight noise, try neatimage:
http://www.neatimage.com

(I'm the URL bunny)
Patrick Porlan said:
Do you know if there is some way to incorporate the information
EXIF 2.2 manually to the files tiff that exports Image Shifter?
Click to expand...
I'll try to add that in a future revision. I agree that this would
be a nice addition.
Patrick Porlan said:
Is there some way to introduce a numeric value for the temperature
of the color?
Click to expand...
No.
Patrick Porlan said:
I have checked with other programs to process files MRW that only
process the files of the D7. The files of the D7i process them
corrupt.
Click to expand...
They haven't been updated yet. It would be good if people could put
a few D7i MRW files on some file server.
Patrick Porlan said:
Why Image Shifter processes the files of the D7 and the D7i
appropriately? Is the program able to distinguish when the file is
of the D7 or of the D7i? Both have different icc. I suppose that
they should be processed in a different way.
Click to expand...
Right! This is very true, and for now Image Shifter is using the D7
ICC profile data. However it produces different colors than DIVU
because I'm averaging processed colors with unprocessed ones in the
output. Again, this is totally arbitrary.

-- Patrick
Click to expand...
--
Peter Marina
 
Look in here!

http://www.aim-dtp.net/aim/calibration/xlprofiler/index.htm

Peter Marina
dream said:
Auke Nauta said:
Please note that every colour device requires a specialised
colour profile, that is no 2 devices are exactly the same.
So when I create a custom profile for my camera, you can use it,
but results will be optimal only on my system.
You probably know all of this...

Greetings,
Auke Nauta
Click to expand...
it would be so nice of you to tell us how to make our own icc profile.
Click to expand...
--
Peter Marina
 
Peter Marina said:
That snapshot must be as
unprocessed as possible and gamma has to be 1.0. Is it possible to
do so with iShifter?
Click to expand...
If you uncheck everything in the adjustments menu and set gamma to 1.0 in the levels dialog that should be OK. At this stage, only Bayer interpolation and crude color-balancing has been done. This color balancing is simply scaling of the r,g,b values by three constants provided in the MRW header, and without this step the image would be greenish.

-- Patrick
 
You must log in or register to reply here.

Keyboard shortcuts

f
Forum
m
My threads
Mobile site
About
Editorial content
Cameras & Lenses
Community
All content, design, and layout are Copyright © 1998–2025 Digital Photography Review All Rights Reserved.
Reproduction in whole or part in any form or medium without specific written permission is prohibited.
When you use DPReview links to buy products, the site may earn a commission.
©GPS Media - Guides, Products, Services.
Back
Top