Go Back   Steve's Digicams Forums > Software > Editors (Photoshop, Vegas, Final Cut Pro, Kdenlive, etc.)

Thread Tools Search this Thread
Old Jun 30, 2009, 12:37 AM   #1
Junior Member
Join Date: Jun 2009
Posts: 3
Default need help..

Iam Ronny. Iam new to this forum. I need some help. Iam a student of "Image Processing". Iam need to work on Bayer CFA Images.
Can anyone suggest, how to get Bayer CFA Images(I need some sample pics). Is there any conversion process(programming or software), i.e. for converting Ordinary color Images to Bayer CFA images.
I dont have any Digital Camera.

Also, how can i get RAW Images in internet? I need some sample pics..

Please help me. i need these images urgently.

Thanks n Regards,
ronny123 is offline   Reply With Quote
Sponsored Links
Old Jun 30, 2009, 5:46 PM   #2
Join Date: Jun 2003
Location: Savannah, GA (USA)
Posts: 22,378

Welcome to the forums.

I'm not sure how to even begin responding to your questions. The vast majority of digital camera use a Bayer type Color Filter Array. A digital camera's built in image processing performs the demosaic process so that you end up with a usable image format with red, green and blue values represented at each pixel location (even though each photosite is only responsive to one color, thanks to the color filter arrays being used over modern sensors).

This is a relatively complex subject, with a variety of demosaic algorithms used (and some are better than others, depending on the camera's sensor, filter design, subject type and more).

The raw file produced by a given camera model is somewhat unique (with different byte offsets for information included in the metadata and much more), and a complicated process is needed to convert the image into a viewable format (and different raw converters will have different approaches to giving you the best image possible, with some of them using better algorithms compared to others for this conversion). So, you have to use a raw converter that supports your camera's unique files to convert them into a format that is understood by image viewers and editors.

A camera's image processing when shooting jpeg is making decisions that may not always be what you want. Manufacturers have to decide on what kind of tone/contrast curves to use to make an image that looks good to most viewers, and sometimes they use processing that can be a bit too contrasty (causing loss of detail in shadows and highlights), in order to give cameras owners a more "punchy" image straight from the camera.

Ditto for things like sharpening, which is mostly increasing the contrast at color/brightness transitions in an image using edge detection techniques (which can destroy detail if overdone). The same thing is true for how the camera compensates for the temperature of the lighting you're shooting in (your White Balance settings handle that). You can easily modify things like White Balance later if you shoot in raw (basically, white balance is nothing more than a set of RGB multipliers that are applied during the conversion process).

The camera's jpeg processing is also limited by the speed of the cpu/hardware for image processing built into the camera, as to the sophistication of the algorithms used, since they are trying to process images in a split second between shots. As a result, you can usually do better if you shoot in raw, using a high powered PC to process the data from the camera's sensor, without the limitations of needing to process it in a split second that the camera's image processing has to deal with.

The individual photosites in a sensor are only sensitive to one color each, and with most Bayer Pattern sensor designs, you have twice as many pixels that are sensitive to green. The raw conversion algorithms take the values from the red, green, and blue photosites and combine them via sophisticated interpolation techniques so that all 3 colors are stored at each pixel in order to give you a usable/viewable image.

The raw file has not combined the output of photosites in any way. That's what the raw conversion process does (or in camera processing if you shoot in jpeg). There are a number of different algorithms used, and some are better than others. You can see some of the common ones discussed here (requires Adobe Acrobat Reader):


Here is another document discussing raw conversion techniques:


When you convert to TIFF or JPEG using a raw converter, you've gone through this demosaic process of combining the values from the red, green and blue sensitive photosites so that red, green and blue are stored at each pixel location. Ditto for shooting in jpeg (the camera is performing the raw conversion if you shoot jpeg).

Most raw converters are doing some additional processing, too (sharpening, contrast, etc.). Ditto for in camera processing (which is doing the raw conversion for you).

Personally, I'd save the raw files, no matter your preference in raw converter, as technology is continuing to improve and you may want to reprocess cherished images later using better tools.

For example, beginning with dcraw.c version 7.60, David Coffin started using AHD (Adaptive Homogeniety Directed) for the interpolation algorithms. This was in part due to Paul J. Lee, who collaborated with Hirakawa Keigo (the original co-author of AHD algorithm as part of his Ph.D thesis).

Once Paul fully understood the algorithm, he modified dcraw.c to use it and gave a prototype to Dave. After some comparison studies, Dave was convinced that AHD was superior to VNG and other demosaic algorithms. Dave converted and optimized the prototype created by Paul J. Lee, integrating it into dcraw.c versions 7.60 and later (and he's continuing to refine it). Note that most products using Dave's code allow you to select between multiple demosaic algorithms (since Dave' code still supports older algorithms like VNG).

Products using some or all of David's code to perform the raw conversion benefit from these improvements, while adding their own enhancements and features. Keep in mind that a number of other raw converters use some or all of David's source code for the demosaic part, while adding a GUI front end and more features to augment the conversion algorithms.

See the "Other Raw Photo Decoders" section here:


UFRaw is one popular free product that uses Dave Coffin's code for the demosaic algorithms.

To get a raw file (either native or .dng) into a usable (i.e., viewable or printable image), it needs to be converted to another format that has already gone through the demosaic process (for example, to a TIFF or JPEG image), combining values for red, green and blue at each pixel location. That's where a raw converter comes in (Adobe Camera Raw, UFRaw and many more).

If you really want to understand it, you may want to consider studying Dave Coffin's dcraw.c source code. Dave's a pretty sharp guy (and a really nice guy if you spend some time getting to know him). He's really dedicated to helping others get the most from their camera's raw output. You may also want to take a look at the latest libraw library (which uses a lot of Dave's dcraw.c code).


For raw images you can download to work with, try this site:

JimC is offline   Reply With Quote
Old Jul 1, 2009, 12:38 AM   #3
Junior Member
Join Date: Jun 2009
Posts: 3

Hi Sir,

First of Very Very Thank you, for your good response.

you have given me good information on Camera processing, about different types of images like RAW, JPEG etc, about Demosaicing Algorithms, and also about RAW converters.
But What exactly I need is Bayer CFA Images(which has only one color component(R or G or B) at each pixel), which is to be given input for my project. Cant I get Bayer CFA Images from RAW Images.
Iam little bit weak in Image Processing, as Iam new to this. Just now starting to learn the subject.

Thanks n Regards,
ronny123 is offline   Reply With Quote
Old Jul 2, 2009, 10:35 AM   #4
Join Date: Jun 2003
Location: Savannah, GA (USA)
Posts: 22,378

There are some utilities that let you see the data in a visible way before the demosaic process. I'd try this one. I haven't used it, but I think it can display the raw file as gray scale without interpolation. It looks like it's got a lot of features from what I can see browsing through it's online user guide.


Also, if you look through the interpolation options section of the manpages for dcraw.c, you'll see that you've got a couple of options (-d and -D) to turn interpolation off.


If you look at the Frequently Asked Questions section of Dave's main page, you'll find links to sites that have dcraw.c already compiled for Mac and Windows platforms (and it's already in the repositories for most linux distros, or can be easily compiled from command line).

JimC is offline   Reply With Quote
Old Jul 3, 2009, 12:23 AM   #5
Junior Member
Join Date: Jun 2009
Posts: 3

Hello Sir,

First of all "Thank you" for your Response.
I think these are useful to me. I will check/try these methods.

Thanks and Regards,
ronny123 is offline   Reply With Quote
Old Jul 3, 2009, 9:52 PM   #6
Senior Member
VTphotog's Avatar
Join Date: Mar 2005
Location: Extreme Northeastern Vermont, USA
Posts: 4,240

You may also find additional information and links here:http://members.optusnet.com.au/pszym...r/graphics.htm

The RAW converter is specific to Minolta cameras, but the links, if still valid have some very good information.

VTphotog is offline   Reply With Quote

Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off

All times are GMT -5. The time now is 10:11 PM.