[sane-devel] Getting Clever CAM 360 IS NOW WORKING PARTIALLY - running gphoto2 version 2.1.6

[kilgota at banach.math.auburn.edu]

On Sun, 18 Sep 2005, Martin wrote:

> He are my testing results (see attached jpg file).
>
> - still mirrored

Yecch.

> - looks more grainy
>

That is a side-effect of the colors being wrong.

> 1) The oranges in the actual newspaper turn out as green in the picture (see 
> the backwards $ prices).
> 2) The greens in the actual newspaper (see the backwards "5X") turn out ok as 
> dark grays in the pic.
> 3) The yellows in the actual newspaper turn out as light grays in the pic.
> 4) The blacks are ok.
> 5) The reds in the actual newspaper turn out as green.
> 6) The blues in the actual newspaper turn out as a gray.
> 7) overall less colour shown in this configuration.
>

as I said:

>> I did the following special steps while 
>> converting to PPM format:
>> 
>> Vertical flip (byte-reversal of entire image)
>>
Comment: This reverses everything. So the content of each line is 
reversed, and the line also gets moved.

>> horizontal flip (byte-reversal of each individual row)
>>

Reverses contents of lines a second time. Uneconomical, perhaps. But I did 
it this way for another driver, where 95% of the cameras need the first 
operation, and then 5% need the second, afterwards ...

Perhaps what you need to do is to move just the lines, so you might try 
one of the other postprocessing routines.


Remark: the order in which I did these does not matter. But after I look 
at the source code in camlibs/polaroid it does seem that I am doing this 
differently. I did the vertical and horizontal flip first, and the Bayer 
routine after. The camlibs/polaroid routines say to do the Bayer routine 
first and any vertical and/or horizontal flip after. That _does_ make a 
difference.


>> 
>> and after this I used BAYER_TILE_GBRG
>> 
>> and got a decent photo with the colors only slightly off from the scanned 
>> image.

But I did the operations in the opposite order, as I said. Thus, the entry 
is still off. I got the photo right, but I did it with my operation 
sequence, not with Marcus Meissner's. You need to fit your configuration 
to the way he did it, if you want to use his code.


>>Thus I think your entry is a little bit off and probably you should 
>> change it to read as follows
>> 
>> 
>> {"Clever CAM 360", 0x797, 0x8001, {
>> jd350e,
>> BAYER_TILE_GBRG,
>> &trust350fs_postprocessing,
>> "scope%04i.ppm"
>> }
>> },
>>

Well, you need to change it again. Probably it will be better if you work 
on the color tiling first, as that comes first in the processing of a 
photo, the way it was done here. Probably is BGGR or RGGB, and the 
postprocessing is probably the jd350 postprocessing. But again these are 
guesses and if I had the camera in my hands it would not take long, I 
suspect. I will wager that it should not take you very long, either, just 
to apply trial and error to a short list of possibilities. Might also help 
you if you go and look at code for the several postprocessing routines.


>> Mind you, I cannot test this directly without a camera; Marcus Meissner is 
>> the expert on the polaroid driver. I am not. I can only report what I had 
>> to do to get a half-decent-looking photo out of your raw file. The rest is 
>> speculation, but this is probably a good place to start.
>>

You can read the above paragraph again, if you are not in too big a 
hurry. Just remember that nobody's opinion counts. What matters is whether 
it works or not. "The rest is speculation ..."

Cheers,

Theodore Kilgore