Re: [sane-devel] Converting Raw Image Data to PNG
Hi Daniel, Andy, list, Sorry, am a bit pre-occupied and busy with the upcoming 1.0.29 release :-) Andy Bennett writes: > Hi, > >> Thank you for your responses. I was able to get something! Almost there. >> >> I've attached an image of progress so far. By preapending the >> "magic pnm code" to the top of the data file I was able to get >> the raw scan data to display, but not properly. It seems to be >> separating the images into three sections, seemingly RGB. The >> sideways slant of the image is also a bit off as I couldn't >> specify the width with a double. > > I think you have the dimensions slightly off, but the width (1296) seems > correct. > As you say, the image appears slanted and when I try to convert RGBTest.pnm > to png I get the following: > > - > $ pnmtopng RGBTest.pnm > RGBTest.png > pnmtopng: EOF / read error reading a row of pixels > - > > It looks like there's an extra 5 pixels every 15 lines or so, except for > every 5 set of 15. > > 15 lines fit into 20KiB, but 16 do not. I'm not sure if that's relevant at > all. Perhaps there's some extra framing, checksum or meta data coming in > from somewhere? I had a quick look at the image only but it appears to be image in a format very similar to p where every R, G and B, stands for a single pixel's RGB values. You'll have to rearrange those in your backend. The p stands for padding bytes and need to be skipped when preparing the image data that your backend returns. This kind of image data is(was?) pretty common and you should be able to find usable code in several of the SANE Backends. I know that the epson (and epson2?) backend is able to handle such data. Hope this helps, -- Olaf Meeuwissen, LPIC-2FSF Associate Member since 2004-01-27 GnuPG key: F84A2DD9/B3C0 2F47 EA19 64F4 9F13 F43E B8A4 A88A F84A 2DD9 Support Free Softwarehttps://my.fsf.org/donate Join the Free Software Foundation https://my.fsf.org/join
Re: [sane-devel] Converting Raw Image Data to PNG
Hi, Perhaps just raw raster pixel data? Although the ordering of the bytes and channels might take some figuring. I was able to get the raw scan data to display, but not properly. It seems to be separating the images into three sections, seemingly RGB. The format of PNM (with a colour depth of 255) is a byte per pixel per colour component in the order red then green then blue. So I guess you'd have to buffer each channel as it comes in from the scanner and then interleave them after each page? This page shows you some examples http://paulbourke.net/dataformats/ppm/ The examples are in P3 rather than P6 format which just means the pixel component data is encoded as ASCII numbers rather than binary bytes. See "PPM example 4". The sideways slant of the image is also a bit off as I couldn't specify the width with a double. How are you processing the image data? Can you share a code snippet? Best wishes, @ndy -- andy...@ashurst.eu.org http://www.ashurst.eu.org/ 0x7EBA75FF
Re: [sane-devel] Converting Raw Image Data to PNG
Please note that the image could be truly raw data from the scanners read heads, not like the 'raw' files produced by cameras, which typically contain lots of meta information about the camera and the image. The data you get could be scrambled, interlaced, interleaved, inverted and uncorrected, and (unless it is JPEG) will contain no header at all. Fortunately, you can open such a file with a good text editor, and add a PNM header, then tweak the header until you start to see an image form. The latter part of this presentation I gave a few years ago shows the process I use: https://www.youtube.com/watch?v=pJ60vVOKVr0 allan On Fri, Jan 10, 2020 at 11:05 AM Chris Moller wrote: > > ufraw and dcraw will handle that conversion. > > On 10/01/2020 00:09, dan...@pixelgrease.co wrote: > > Hello, > > I have been trying to get a device running for a few weeks now. After failing > to build a backend I decided to hack together my own LibUSB driver for my > unsupported A8 scanner, given my use case is pretty simple I went in with a > "how hard can it be?" attitude. > > Great success so far. Hasn't been easy but it hasn't been really that bad. I > have managed to ask my tiny A8 feed scanner to scan a 300dpi colour image. I > have been able to save the data to a file (about 8mb worth) but that's where > the fun ends. > > Can't seem to figure out how to get that raw data into a usable format. 2 > questions: > > 1. What format is that raw data likely to be? I naively expected it to be png. > 2. Does SANE have some code that handles that sort of conversion I could have > a look at? Looked around the internet and the SANE docs. Read some things > about frames, but couldn't find much else. > > THANK YOU! > > Daniel > > > -- "well, I stand up next to a mountain- and I chop it down with the edge of my hand"
Re: [sane-devel] Converting Raw Image Data to PNG
Hi Daniel, dan...@pixelgrease.co writes: > Hello, > > I have been trying to get a device running for a few weeks now. After > failing to build a backend I decided to hack together my own LibUSB > driver for my unsupported A8 scanner, given my use case is pretty simple > I went in with a "how hard can it be?" attitude. > > Great success so far. Hasn't been easy but it hasn't been really that > bad. I have managed to ask my tiny A8 feed scanner to scan a 300dpi > colour image. I have been able to save the data to a file (about 8mb > worth) but that's where the fun ends. > > Can't seem to figure out how to get that raw data into a usable format. > 2 questions: > > 1. What format is that raw data likely to be? I naively expected it to > be png. Hmm, more likely to be "raw" data or JPEG. Save the image data you get to file as is and let the `file` command take a stab at it. It should recognize standard image file formats. If it's raw data, you want to note the image width, height and depth of your scan and try prepending a PNM header to the data you got. > 2. Does SANE have some code that handles that sort of conversion I could > have a look at? Looked around the internet and the SANE docs. Read some > things about frames, but couldn't find much else. If it's raw data, you can very likely just pass it to a SANE frontend as is (assuming 1, 8 or 16 bits per pixel and telling the fronted that). For JPEG, there are a number of backends that already handle that. Just run `git grep -l jpeg -- backend` to find them. For PNG, you may want to check the brand new escl backend. Hope this helps, -- Olaf Meeuwissen, LPIC-2FSF Associate Member since 2004-01-27 GnuPG key: F84A2DD9/B3C0 2F47 EA19 64F4 9F13 F43E B8A4 A88A F84A 2DD9 Support Free Softwarehttps://my.fsf.org/donate Join the Free Software Foundation https://my.fsf.org/join
Re: [sane-devel] Converting Raw Image Data to PNG
Hi Daniel, On 1/10/20 8:09 AM, dan...@pixelgrease.co wrote: 1. What format is that raw data likely to be? I naively expected it to be png. 2. Does SANE have some code that handles that sort of conversion I could have a look at? Looked around the internet and the SANE docs. Read some things about frames, but couldn't find much else. Raw data is likely to be just a matrix (2D array) of pixels. One byte per pixel in grayscale mode and 3 bytes per pixel in color mode. In SANE world, there are drivers, called "backends" and client programs, called "frontend". Backends are dynamically loaded libraries, such as sane-pixma (/usr/lib/sane/libsane-pixma.so.1) and so on. Frontends are programs, line simple-scan, scanimage, xsane or even gimp. Frontend always returns picture in raw format, even if it comes from device in some another format (say, PNG or JPEG). This is a responsibility of frontend to convert the image into something useful, like JPEG or PDF file. So if you are about to write sane backend, this is not your responsibility to convert the image into PNG, but if image comes from device as PNG, this is your responsibility to unpack it. -- Wishes, Alexander Pevzner (p...@apevzner.com)
