PV2Fetch releases
PV2Fetch 0.1 initial release is out. ( http://sourceforge.net/project/showfiles.php?group_id=134210&package_id=169980 ) It's a Windows program, though I think it would be pretty easy for someone to split out the image-manipulation code and create a similar utility for Linux. Here's the ReadMe so you guys can see what it's about... Description PV2Fetch automatically downloads RAW images from SMaL chipset based cameras, and then creates jpeg images from those RAW files. Just like the commercial SMaL RAW drivers, it processes the heck out of the images to improve their percieved quality. Currently, the PV2Fetch applies the following filters: -white balance -variable gamma correction based on level of darkness in image -saturate -unsharpen mask and noise suppression If you prefer to carefully retouch each image, PV2Fetch may not be for you - you're likely better off with RAWGL in that case. I like to think of PV2Fetch as the Free Software answer to the commercial SMaL drivers. Installation You need your computer setup with libusb before you can use PV2Fetch. It's recommended that you get PV2Tool working first before you try PV2Fetch, since it gives more feedback on errors. One you have that done, just drag and drop the PV2Fetch executable that matches your libusb version into your image directory. Usage Mode 1 - RAW FILES ON CAMERA Double-click the executable. PV2Fetch will connect to your camera and offer to download the pics. The download and coversion is automatic. Afterward it will offer to format the camera. Mode 2 - RAW FILES ALREADY IN DIRECTORY Drag-and-drop a single RAW file onto the executable, and it will convert it into a jpeg. License It's GPL software folks. Read GPL.txt for the stimulating details. Copyright Forkboy, Sailpix, Dave Coffin, the libusb team, the libjpeg team, and yours truly, BillW. Thanks Special thanks to Sailpix - without his tireless disassembly and analysis we wouldn't have any of the Free RAW decoders. Additional thanks go to Binaryweaver who helped with -*test*-('")ing!

Great Program!!!! This is exactly what I needed for the kids. For version .2 could you please consider: - add a configuration parameter that the program parses at start up so it can determine if it should just land the pictures or if it should make a directory off of the 'Start in' setting of the short cut? This one feature would really help the young ones keep their pictures organized. (I just don't have time to install Borland's compiler right now! If it was MSVC I would knock it out.)

I don't know jack about coding, so I figured I'd request an additional feature.. Perhaps on the question that asks if you want to extract the pictures, can you put an option to delete the .RAW files? I would guess a checkbox or something would be easiest. Otherwise just a seperate question.

Bill, I haven't looked at yer new program yet... but concept sounds dead on. What's the build environment? (or, how easily could I change it over to Visual Studio build so that Create987 can help code?) Bupkis - in another thread you wrote: Only thing is tho, you can definitely tell the difference in pictures from the FF2 drivers. Sometimes the pv2fetch ones are better, sometimes not. Can you highlight/provide/upload/identify some specific examples of good and bad processing - preferrably with .RAW and .JPG samples? It is likely that we can add some more logic in the photo processing sequence, but first we need a better picture (pun intended) of what's "working" and what is "broken". This is, of course, somewhat subjective but we'll need to start somewhere.

Well, here's a couple examples that I just now took... Guy with FF2 drivers Guy with PV2Fetch app The FF2 driver pic looks a little more detailed, a little closer to flesh tones, and a slight bit sharper. But in this example.. Palm with FF2 drivers Palm with PV2Fetch app The FF2 driver pic is much brighter. Seems like the PV2Fetch pic is more easy on the eyes. So far, when photographing people, the FF2 driver seems more ideal. In other situations though, the PV2Fetch one is a bit smoother. I can upload the RAW files someplace if you guys want.

Dragging .raw files onto it doesn't work. It says it converted it, but no sign of the JPG.

Cool comparison bupkis - I'd love it if you could either post the RAWs or send them to me (billwiley777 at gmail dot com) I'm getting close to a 0.2 release. I'll post a directory full of comparisons including yours, if that's ok, so people know what to expect. Neat idea Create987 - I'll see what I can do once I've gotten the processing as good as it can get.

PV2Fetch 0.2 is out. ( http://sourceforge.net/project/showfiles.php?group_id=134210&package_id=169980 ) Changelog says it all... 0.1->0.2 11/21/05 -changed slight-sharp filter to unsharpen mask. -changed luminosity curve. In the darker end it's a more traditional logarithmic curve and toward the brighter end it's linear. -increased saturation. -clipped extra image border.

Here's the comparison shots between ff2, pv2fetch v0.2, and in some cases CVS processing thanks to brite_eye. http://camerahacking.mine.nu/phils/demanufacturing/pv2/www/billw/pv2fetch/

Wow, the cv2 ones really do look the best. How do we get the software they use? :) hehe I'll send those RAWs to you asap.

The calib1/cvs (not cv2) pics have the most accurate color. For further reading on my experiements: forumer.com/viewtopic.php?t=386" target="_blank">http://camerahacks.10.forumer.com/viewtopic.php?t=386 forumer.com/viewtopic.php?t=387" target="_blank">http://camerahacks.10.forumer.com/viewtopic.php?t=387 http://www.photo.net/bboard/q-and-a-fetch-msg?msg_id=00CqSe Great job BillW - I think your processing is comparable to FF2, but I still like the realism that CVS achieved (my basement wall is dirty but does not have eiry bluegreen cast). An interesting experiment would be to actually take a pic of color chart, develop picture and then take a pic of the color chart next to developed photo - that would eliminate confusion resulting from poorly calibrated monitors. IMG_16 red curtain reveals the grea-*test*-('") differences IMO.

I keep calmly asking professional photographers if they have a Gretag Macbeth color standard that I can borrow to photograph with a SMaL camera. General answer seems to be "I know what that is, but I don't have one..." Some photographers (pros, even) don't seem to be _that_ concerned with color correction - but, they have nice high-end cameras and digital photo labs that will fix that stuff for them. If anyone has one of these charts, it would be most helpful if you could shoot a full-frame picture with a PV2 camera in sunlight at high noon. From the results I think we can do some calculations to make a "transfer function" for data from these cameras. Or at least, that's my theory for the moment...

I was thinking a bit about color calibration and something brite_eye said to me about different lighting conditions requiring different calibrations. Perhaps they took one calibration shot indoors and one outdoors and averaged the results together to achieve the best average performance. Interesting note on the basement curtains pic brite_eye. Both pv2fetch and ff2 have curtains with too little blue and walls with too much blue. Clearly some heavy calibration adjustments going on in the CVS service! Coolty, sorry but I have to ask - are you sure the new file didn't show up at the bottom of the list? (ie. not alphabetical). If not, then send the raw to the mail address I provided previously and I'll see what's up.