First image from naked CMOS sensor ...Kinda...
After much swearing and gnashing of teeth 8) I was able to read something that roughly resembles an image out of the CMOS sensor with a microcontroller. Since it's just sitting on a little board and there isn't room to really put the lens on, I just half-covered it with a piece of paper at an angle so I could tell if the data was total garbage or not. The setup Image read out First thing you'll notice is that it's noisy as hell, longer than 864 px vertical (I don't know what's going on there, but see next point), and a bit scrambled up in terms of both the vertical line positions and intensity. Intentional obfuscation by SMaL, or just a confused sensor, not being properly reset or clocked? I'm leaning toward confused sensor for the time being, since I still don't really know what's happening and can't reliably reproduce it. If there's a specific reset sequence that puts it in a known state and/or makes it reliably capture an image, I haven't found it. (Some seemingly regulated analog voltages *should* come up on most of pins 17 ~ 24, but it doesn't always happen. Powercycling a few times seems to work, for now.) I'm using pin 10 from the imager as feedback to tell when the next line of valid data is coming (from clocking a 'line clock' on pin 16, and 1288 'byte clocks' on pin 12 for each line clock). What I'm seeing is the first line clocked out is valid, the 4th, and then every 96th after that. (Note: 96 divides evenly into 864, the total V size.) Pin 10 raises during the 95th line (a repeating garbage byte on D<7..0>) and lowers again at the start of the 96th, where the data is valid. All lines in between are read as 0x00s. Anyway, some progress is better than no progress. I'll poke at it a little more :)

Very interesting. I have been bothered for a long time about SMaL ultrapocket 3 specs of 1280 x 960 not matching these cameras and have concluded that someone screwed up in LCD design resulting in need to chop off bottom of picture to make room for remaining message. Note that SMaL ultrapockets 3,4,5 all have a 4:3 ratio of Vert:Horz. A ratio of 4:3 appears to be an industry standard. Keep us posted. I am still curious about "8 or 12-bit digital (selectable)" output claim on SMaL's website. I managed to remove glass cover on one of these imagers and have started trying to scrape off bayer - without much luck and no way of verifying since the camera has been fried. Still hoping for a resurrection (no luck using billw's method yet).

Drmn4ea, If you want -*test*-('") imagers please post a note - some firmware fried cameras are being junked.

I've interfaced some tektronix sensors in the past and they often had a handful of dead pixels before and after each line, so the actual width may be more than 1288. It was also quite common that they had a few dead lines before and after the picture. Though, looking at your picture I'd say it's more complicated than just an modulo error. It looks indeed like the sensor isn't liking the way it's being handled.

Drmn4ea, If you haven't read Fife's thesis recently, I suggest reading section on bit planes where LSB half of row is sent before MSB half. I believe that is why your image appears sliced and diced. I just updated links with his thesis and patents today.

I have this project 'on'hold' for the time being, hoping that I finish up the PIC USB stuff first. Reading via RS232 at 38400, it takes about 10 min. to find out if it worked or not...

Copy of Drmn4ea's post in lens adjustment topic of how to forum: I have experimentally-determined pinout info (incomplete) on it here, but unfortunately haven't been able to find any real specifications on it. @brite_eye: you actually removed the glass and scraped on the chip? Is it still in-camera, and can you take pictures on it? (I'll be a little surprised if the chip still returns data :-) May want to -*test*-('") it periodically, before spending too much time at once scraping at a dead imager.) I'm also wondering if the controller SMaL put into the PV2 is even capable of taking advantage of Autobrite stuff, if any. From what I can tell, the data lines to the imager are input-only (not bidirectional...evidence is the pullup resistors selectively placed on some of them, maybe corresponding to a single command byte to the imager? If the ASIC could drive those values itself, they could save a couple pennies and omit them), so I can't see how the ASIC would e.g. adjust gains or other parameters on the fly. I have stopped scraping and am hoping I can resurrect and take a picture on my fried Ritz 6410. I am now wondering more about Drmn's comments on imager and ASIC not implementing Autobrite. I have been unable to find any references outside of our hacking that ties Autobrite to CVS cameras. However Radioshack cameras FF2 and FF3 do advertise Autobrite on packaging and PDF document on CD. If anyone locates an outside Autobrite reference specifically related to CVS or Pure Digital cameras please post here.

What do you mean by "outside"? My Wolf/Ritz 6410 one-time-use camera displays the AutoBrite patent number on it's startup screen. So... I assume that the technology is inside there compressing the light levels somewhere.

An eye for detail, not unexpected from a sailor! I never saw it before, but in very tiny letters it passes quickly on my 6520. By outside I meant any autobrite evidence besides these forum discussions. Maybe Drmn4ea needs to perform more probing or could logic be embedded in cmos imager?

Brite eye getting to nitty gritty
Very Interesting Brite eye ! ------------------------------ "I have stopped scraping and am hoping I can resurrect and take a picture on my fried Ritz 6410. " -------------------------------- WOW :lol: . This is what I am most curious about. What does you Get in the RAW with all the analog stuff out of the way. What does digital UV and IR like in the RAW??? This is a big deal IMO. Thx for the info and work todate.. Please Just keep on keepin on. There are no CVS stores here but there is much I can learn until I get my own CVS s. In the mean time I can live vicariously from the posts.

It's entirely possible that Autobrite functions are incorporated on the imager in some semi- or fully-automatic way... if so, the ASIC would have no need to write data back to it. Then again, in a brief side-by-side I did between the old Blue (Sunplus) and PV2, the PV2 exhibited some very visible saturation in bright areas, something the Autobrite technology is supposed to prevent (note the white pants of the guy in the background, and the windows at top-right). I will be probing some more on the imager, when I get my chips back (temporarily borrowed for work)..

Although there is no solid proof one way or the other yet, I believe autobrite is in-imager and automatic. Like brite_eye, I am less afraid of sun-behind-subject with pv2 and ff2. Those pics seem to generally turn out ok, which is more than I can say for my other digital cameras. Drmn4ea, I think the oversaturation is probably due to over-agressive gamma correction in the driver. It would be interesting to see the same pic developed with daBass' raw viewer. If the windows appear oversaturated *without* gamma correction, then we know the imager was at fault. Do you still have the raws?

Based on Drmn's pix, the micron imager with larger lens looks much better on an overcast day. Clearly the additional blur in second shot is due to cat shake during a longer exposure time. I suggest avoiding dark shots and trying some directly at the sun. Note that even FF2 and FF3 have a lens that while bigger than new SMaL 1.3 is smaller than old pv1 Dakota (I wish these lenses were standard sizes so I could order a telephoto). Here again is my brite_sunny day comparison discussion on photo.net clearly showing much more detail in shadows without losing detail in sky: http://www.photo.net/bboard/q-and-a-fetch-msg?msg_id=00BijK The discussion also clearly shows that those owning expensive cameras are disbelievers. IMHO the jury needs more evidence and I am hoping that removing bayer and using a better lens will better demonstrate autobrite advantage. I also think my CVS SMaL autobrite brite_eye captures a notably greater dynamic range from bright flash to reflection of camera in dark pupil. http://www.photo.net/photodb/photo?photo_id=3080276&size=lg (CVS Red 27) http://www.photo.net/photodb/photo?photo_id=2834686&size=lg (old dakota)

On the lenses, it's an M9x0.5 thread format. Checkout http://www.mars-cam.com/optical.html or http://www.mars-cam.com/lenses/ccd_cmos/assembly.html or type in "board lenses" in Google and a lot of time looking.

CameraWorld now owned by Ritz and the place I bought both my original Dakota blue and recently the Red PV2. Has both Telephoto & wide angle lens for simple Digital cameras. They run $25.00 each apox. These are made for small camera lens and I am pretty sure they will fit the original dakota but not the PV2/CVS. They have two prongs that push against the inside of the plastic funnel shaped extension that is over the lens in order to stay connected. There is like a sliding push in on each side thing that lets you put it on over the lens. If I remember correctly they fit lens from 9 -24 mm . The Dakota original funnel is apox 19 mm max size . The PV2/CVS doesn't have anything to grab on to but perhaps it could be modited to fit. I may buy one at some point. PS. It may have been close up and wide angle lens rather then telephoto . I'll have to go down and check one day. They don't have them online .