Scanner Camera at Pikes
June 7, 2008 2:26 pmTook the scanner camera down to Pikes market during one of their festivals. This was the first time I had the camera outside, and the results were encouraging. You can see all the pics here:
A couple things to note:
I turn off course, fine, light and dark calibration. I use the alt calibration option. I think the per-pixel calibration that is stored in DRAM during the alt calibration is a bit off for the pixles near the top of the image. for example, see this image:
You can see the light band of pixels at the top of the image. This gets significantly better when stiwching from grayscale to color images. The color images come out very very grey. There are two things that I want to look into further:
- Better understanding/knowledge of the calibration. I think this will significantly improve image quality.
- Ensure that I’m operating the scanner ship in 14 bit mode, and that I’m getting out all 14 bits. If the chip is in 14 bit mode then there is no calibration data stored in DRAM (the space is used for pixel data.
Matt
6 Responses to “Scanner Camera at Pikes”
So, I was attempting to decompile the sigma sd14 firmware, seeing as that I own one, and that you are farther along than me, I thought I might ask a few questions.
But,
as you notice, I am posting this in your scanner camera category, considering that I have built one from a canon scanner
http://spigget.deviantart.com/art/Big-Camera-number-3-112351759
and it takes nice photos
http://spigget.deviantart.com/art/Dust-to-Dust-93220521
Hey, thanks! What scanner did you use? What driver/software are you using? I’ve made some tweeks to mine, and the image quality is improving:
http://www.mattmerry.com/scanner/index.html#images/baker/baker0001.jpeg
and:
http://www.mattmerry.com/scanner/index.html#images/baker/baker0017.jpeg
Its still a work in progress though.
Are you using color filters and multiple scans to get the coloring effects, or is that straight out of the scanner?
I have a box of 6X6 inch drop in filters that I use.
http://spigget.deviantart.com/art/Filter-Box-113269480
they slide into a filter holder in the front. The front filter holder is not shown in that image I sent you of the big camera.
I also have a few 82mm thread on filters as well
>950nm IR
>850nm IR
>760nm IR
>720nm IR
>650nm Red+IR
>500nm Yellow+IR
Blue channel
Visible (490-650nm) -> Green channel
Infrared (820-950nm) -> Red channel
and there are seven scans I took for that image, in this order
Ultraviolet
Visible
Infrared
Ultraviolet calibration
Visible calibration
Infrared calibration
Dark frame
I take all of these when I take the photo because the sensor changes sensativity depending on the spectral band used, temperature, humidity, etc.
The focal length on most lenses will vary across spectral bands, so each image is refocused foroptimal results.
Usually I pre focus, and pencil in marks on the focus slider so I can capture the 3 images as close together as possible to reduce changes in the scene.
The Canon scanner you are using has a monochromatic strip sensor, so you need to capture 3 images for color and use 3 corresponding color filters. Typically you want these to closely match the spectral response of the human eye, but that’s up to the artist.
You could synchronize the RGB LED output in the scanner to red green and blue studio lights to achieve color, but I’ve never tried this, and I would imagine it would be quite difficult.
The scanner in my older camera is a Canon LiDE 70, which may be the same one you use.
I use Hamrick’s VueScan for my captures
http://www.hamrick.com/
Hope that helps a bit.
and I would be glad to answer any other questions you may have, as I’m sure you will probably have a few.
So, I’m trying to modify the firmware on my SD14 so that it saves the original sensor data to the .raw file instead of running it thru color space conversion and noise reduction.
I thought I’d ask, and I’m sure I could wade thru your notes, but you might be able to give me some pointers on where to look and what to try.
Interesting. I intended on trying a 3 scan with RBG filters this past weekend but managed to leave the blue filter at home, so I did not run the trial. I’ve noticed the sensor is highly sensitive to IR so I’ve been looking around for an IR+UV cutoff filter to try. A lot of the ugliness of my initial scans came from IR light inside the scanner and the fact that the plastic back was not opaque.
Did you make any physical modifications to the scanner?
I’m working on modifications to Sane the goal being to develop a frond end that is more “photographic” feeling as opposed to operating like a scanner. Correspondingly, the backend will need some work as well.
From what I can tell in the SD14 firmware, there are the following stages to the Sigma/Foveon image processing pipeline:
AutoAdjust
LinLUT
ColumnFilter
ColumnFilterSegCapacitance
SpatialGain
ISOGain
DriftCorrection
FlareCorrection
SatDetect
BadPix
RedSharp
NeutHilitePixel
NeutHiliteNeighbors
NeutHiliteSmoothNeighbors
Despeckle
SpatialDQ
ChromaBlur
ColorDQ
CorrectionMatrix
ToneGamma
Sharpen
This may or may not be complete, it comes from the debugging information in the strings. I’m just getting to the point where I’ve identified the individual binary sections in the firmware. I’m also able to extract (and thus examine) the FPGA programming. I’m not to a point where I can modify and run my own firmware. Its slow going - I work on it when I have time.
I did modify the scanner prior to use.
The light guide, the tri color LED, and the lens array have been removed as to let the light focus directly onto the sensor strip. I also trimmed the black plastic housing for the sensor strip as to allow light from sharp angles to reach the sensor.
http://www.usa.canon.com/consumer/controller?act=CanonAdvantageTopicDtlAct&id=6167
on the next one I make, I will remove the glass as well, because I get a lot of problems with dust on the glass. If there is dust on the sensor, it is removed in the calibration, but if it’s on the glass, it shows up as specks.
if you want a good IR blocking filter that closely matches the human eye response, you should try LDP’s X-NiteCC1. it has a nice taper and it does let UV thru but combined with the spectral response of silicon, it works well because silicon is not very sensative to UV.
http://www.maxmax.com/aXRayTechnical.htm
you may also try hoya.
http://www.hoyaoptics.com/color_filter/color_compensating.htm
they have a similar transmission spectra. The good thing about hoya is that you can get the filters in 6.25X6.25 squares. I don’t know if you need anything that big, but I shoot with large apeture sizes to get shallow depth of field. Hoya is where I got my UV filter.
a little bit on spectral photography
http://spigget.deviantart.com/art/Spectral-Series-Number-B-113065418
I do all of my de-line and calibration correction in Photoshop and it would actually be nice to throw away the software calibration data, or zero it out because it is designed for scanner operation, not in the field photography. A lot of the post processing headache comes from removing the software modifications to the raw scanner data, once that is done, then I use my calibration frames to remove the lines in the photos.
I guess as far as the SD14 firmware goes, whenever you can figure out how to skip some of those steps that you listed, that would be awesome, but you probably have a real job like me, so whenever you can.
I would tinker some, but the most programming I have done is on a TI89 graphing calculator and it’s pretty much confusing as heck to me, but I would like to learn when I have time someday
so, I thought of something today.
What filters do you use for your red green and blue primaries?
Care to comment?