Olympus-OM
[Top] [All Lists]

[OM] Re: E-3 BAYER PATTERN

Subject: [OM] Re: E-3 BAYER PATTERN
From: David Irisarri <div2000@xxxxxxxxxxxxxxxxx>
Date: Fri, 19 Oct 2007 00:11:50 +0200
Hi Winsor,

If you download a pdf brochure from KODAK called "KAF-5101CE.pdf". You 
can clearly see that in page 12, it is explained that there are two 
different types of greens and this is why David Coffin decodes ORF 
images as four color RGB images. If think is clear. If E-1 bayer pattern 
had only three colours, they could be decoded with AHD as three color 
RGB raw images without maze pattern.

Please have a look to it.

Regards,

Dave





Winsor Crosby wrote:
> Besides the announced Kodak array with some pixels not color  
> filtered, the only alternative color filter patterns for sensors I  
> have heard about are the CMY arrays used previously on Coolpix  
> cameras and the Sony RGB-E filter which added a turquoise element to  
> RGB. It is the only 4 color array ever used as far as I know.
>
> RGB and RGBG are commonly used interchangeably to describe the Bayer  
> array. The second G is just to point out that there are two green  
> sites for each red or blue site in the standard Bayer array. So I  
> don't know what you mean by a 4 color Bayer array, nor could I find  
> anything about it using google. Maybe you know of a place that  
> explains it?
>
>
>
> Winsor
> Long Beach, California, USA
>
>
>
>
> On / October 17, 2007 CE, at 9:50 PM, David Irisarri wrote:
>
>   
>> Hi Winsor,
>>
>> As you know is very difficult to achive perfect color rendition with
>> imaging sensors and a truly color engine inside the camera would be  
>> very
>> expensive. Despite TruePIC III processor corrects color, high-end
>> profiles will work much better.
>> Olympus decided to use a KODAK sensor with RGRG bayer pattern because
>> improves color rendition but you get lower quality in terms of
>> resolution. This is why many reviewers in the past said Olympus E-1
>> resolution was a little bit lower than five megapixel.
>>
>> You probably have seen that many raw converters work pretty bad  
>> with ORF
>> images because they are using demosaicing algorithms not optimized for
>> four color RGB. A labyrinth maze pattern appears over the image and
>> output quality is very low. David Coffin, has a very good utility that
>> works with four color RGB images, using a demosaicing algorithm called
>> Variable Number Gradients. This utility is called DCRAW.
>> If the camera has a RGB typical bayer patter, it can be used another
>> algorithm that outputs higher quality. It is called AHD Adaptive
>> Homogeneity Directed.
>>
>> So...with four color rgb bayer filters a blur is requiered to avoid  
>> this
>> pattern:
>>
>> http://cybercom.net/~dcoffin/dcraw/ahd_maze.png
>> <http://cybercom.net/%7Edcoffin/dcraw/ahd_maze.png>
>> http://cybercom.net/~dcoffin/dcraw/vng_grid.png
>> <http://cybercom.net/%7Edcoffin/dcraw/vng_grid.png>
>>
>> Sometimes DPREVIEW under camera specs, highlights if bayer pattern is
>> RGB or RGBG but I haven´t found anything according to E-3.
>>
>> Regards,
>>
>> Dave
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>> Winsor Crosby wrote:
>>     
>>> If you are talking about the new Kodak filter pattern where one of
>>> the Gs is gray, I think you probably have your wish. Probably the
>>> sensor is Panasonic as the past several Oly cameras have been and
>>> Kodak, in their announcement, said video first anyway.  I am puzzled
>>> by your statement though because I would have thought the new Kodak
>>> filter would greatly improve the noise floor in physically small
>>> sensors and it had nothing to do with the E-1. What problem was there
>>> with the E-1 filter?
>>>
>>>
>>> Winsor
>>> Long Beach, CA
>>> USA
>>>
>>>
>>> On Oct 17, 2007, at 9:29 AM, David Irisarri wrote:
>>>
>>>
>>>       
>>>> I have been reading everything related to E-3 the whole day and I
>>>> cannot
>>>> find if E-3 is using RGBG or just only RGB bayer pattern filter. If
>>>> TruePIC III processor is able to correct color properly coming from
>>>> E-3
>>>> sensor, I would prefer RGB just only; otherwise we will have the  
>>>> same
>>>> problem as E-1.
>>>>
>>>>         
>>>
>>> ==============================================
>>> List usage info:     http://www.zuikoholic.com
>>> List nannies:        olympusadmin@xxxxxxxxxx
>>> ==============================================
>>>
>>>
>>>
>>>       
>> ==============================================
>> List usage info:     http://www.zuikoholic.com
>> List nannies:        olympusadmin@xxxxxxxxxx
>> ==============================================
>>     
>
> ==============================================
> List usage info:     http://www.zuikoholic.com
> List nannies:        olympusadmin@xxxxxxxxxx
> ==============================================
>
>
>   


==============================================
List usage info:     http://www.zuikoholic.com
List nannies:        olympusadmin@xxxxxxxxxx
==============================================

<Prev in Thread] Current Thread [Next in Thread>
Sponsored by Tako
Impressum | Datenschutz