I've push a patch that reverts the whole set of broken/non-working saturation patches. We should now be back to were we were prior to this. I think having saturation and hue would be nice but it needs to be done correctly and tested first before pushing.
On Tue, Mar 17, 2009 at 4:39 PM, Brian Johnson <[email protected]> wrote: > Oh that is because the hue that we are starting with is different then > normal so your colors are correct. > > On Tue, Mar 17, 2009 at 4:27 PM, Stefan Krastanov > <[email protected]> wrote: >> OK. But why it's only blue? Maybe I was not clear when reporting - the >> saturation script is changing the picture from blue(vivid blue but still >> only blue - no other colours) to b/w to blue again. >> >> >> 2009/3/17 Brian Johnson <[email protected]> >>> >>> What you describe for saturation is what one would expect. Saturation >>> is simpy the color intensity. >>> to start with you are aproaching a zero level saturation so your >>> colors should be getting duller and duller until you hit zero at whcih >>> point you have your black and white image then they start moving away >>> from zero at whih point you should notice your colors get more and >>> more bright and vivid. >>> >>> On Tue, Mar 17, 2009 at 4:01 PM, Stefan Krastanov >>> <[email protected]> wrote: >>> > Thanks - I should really first read the pages on googleGroups. Here are >>> > the >>> > results: >>> > >>> > running sats: >>> > I had picture cycling trough blue-b/w-blue - not exactly saturation >>> > >>> > running hues: >>> > cycling trough red-violet-blue-violet-red-orange-green-blue >>> > I don't know if this is the expected order but it's working. The real >>> > picture was somewhere around orange. >>> > >>> > The picture was in good quality - It was not washed in one colour. >>> > >>> > All that on 0C45:6270 MT9V011 >>> > >>> > Regards >>> > Stefan >>> > >>> > 2009/3/17 Brian Johnson <[email protected]> >>> >> >>> >> By default our driv er is compiled with debugfs support. however you >>> >> probably have to mount the debugfs filesystem try the following >>> >> # sudo mount -t debugfs none /sys/kernel/debug/ >>> >> >>> >> On Tue, Mar 17, 2009 at 3:31 PM, Stefan Krastanov >>> >> <[email protected]> wrote: >>> >> > I checked the ruby script but my kernel/debug directory is empty. I'm >>> >> > searching in the docs for how to compile the module with debug >>> >> > options(or >>> >> > whatever I must do) but if you can save me the time I'll be grateful. >>> >> > >>> >> > There are no problems with saturation (as it's non-existent ;) in the >>> >> > version from Mon Mar 16 00:31:19 2009. But I noticed that contrast >>> >> > bar >>> >> > in >>> >> > v4l2ucp is changing in fact the brightness (and just a bit of the >>> >> > contrast). >>> >> > But it seems that at the moment you are working on the saturation, so >>> >> > I'll >>> >> > remind of that later. >>> >> > >>> >> > All that on 0C45:6270 MT9V011 >>> >> > >>> >> > Regards >>> >> > Stefan >>> >> > >>> >> > 2009/3/17 Brian Johnson <[email protected]> >>> >> >> >>> >> >> Actually I think it would be better to just git revert the entire >>> >> >> saturation commit and redo it with a working commit latter instead >>> >> >> of >>> >> >> just commenting out that line. >>> >> >> >>> >> >> As far as i know the only two broken sensors are the mt9m001 and >>> >> >> ov7670 both i believe give valid pictures but need some tweaking for >>> >> >> color adjustment. >>> >> >> >>> >> >> >>> >> >> On Tue, Mar 17, 2009 at 12:23 AM, JoJo jojo <[email protected]> >>> >> >> wrote: >>> >> >> > >>> >> >> > Hi Brian >>> >> >> > >>> >> >> > Agreed, lets block all other commits before this weekend, >>> >> >> > Guys please post/queue your patches for review this weekend. >>> >> >> > >>> >> >> > Blocking saturation should be simple enuff, maybe commenting >>> >> >> > >>> >> >> > >>> >> >> > >>> >> >> > http://repo.or.cz/w/microdia.git?a=blob;f=sn9c20x-usb.c;h=4eab2d57558ed87cc0eb788066e4acf97f17d123;hb=0963ed513d4f2abff3469292825f971445402392#l729 >>> >> >> > >>> >> >> > Also we need to know which webcams are broken, is it only 6270's ? >>> >> >> > ('coz i seem to remember some sensor color tweaking here >>> >> >> > >>> >> >> > >>> >> >> > >>> >> >> > http://repo.or.cz/w/microdia.git?a=commitdiff;h=e22293994c6019579e24d789be203729699306dc >>> >> >> > ) >>> >> >> > >>> >> >> > And hey, its nice to see people are still tracking/testing >>> >> >> > master/HEAD >>> >> >> > ;-) >>> >> >> > We'll arrange for a more through weekend testing this coming >>> >> >> > friday. >>> >> >> > >>> >> >> > -JoJo >>> >> >> > >>> >> >> > On Tue, Mar 17, 2009 at 9:28 AM, Brian Johnson <[email protected]> >>> >> >> > wrote: >>> >> >> >> >>> >> >> >> Saturation is supposed to change the intensity of a color so this >>> >> >> >> patch is not right as well, Also while hue is supposed to change >>> >> >> >> the >>> >> >> >> actual color this is wrong for that as well since it should not >>> >> >> >> wash >>> >> >> >> the entire image in a particular color like this patch does. >>> >> >> >> >>> >> >> >> I've uploaded for reference an archive containing two data files >>> >> >> >> for >>> >> >> >> both saturation and hue plus a ruby ruby script that can feed >>> >> >> >> values >>> >> >> >> from one of the data sources through the debugfs to show a >>> >> >> >> gradual >>> >> >> >> progression of changing colors. The data values i use for hue and >>> >> >> >> saturation come from the values used under windows. Run this file >>> >> >> >> using the following command, where <datafile> is wither hues.yaml >>> >> >> >> or >>> >> >> >> sats.yaml >>> >> >> >> # ruby ./color.rb <datafile> >>> >> >> >> >>> >> >> >> link: >>> >> >> >> http://groups.google.com/group/microdia/web/color_ref.tar.gz >>> >> >> >> >>> >> >> >> My understanding of how the bridge handles hue and saturation is >>> >> >> >> that >>> >> >> >> you are basically modifying the coefficients for the bridges >>> >> >> >> color >>> >> >> >> matrix. The color matrix is the set of coefficients used to >>> >> >> >> convert >>> >> >> >> the RGB data provided by the sensor ot YUV color-space. >>> >> >> >> Saturation >>> >> >> >> should just be a matter of proper scaling of the matrix values, >>> >> >> >> with >>> >> >> >> hue involving rotating the values using sin and cos functions. >>> >> >> >> >>> >> >> >> The bridge appears to use two bytes per value with the second as >>> >> >> >> far >>> >> >> >> as i can tell either 0 or f which likly is used tell the sign of >>> >> >> >> the >>> >> >> >> first byte. >>> >> >> >> >>> >> >> >> Since both saturation and hue, as well as contrast and >>> >> >> >> brightness, >>> >> >> >> play with the values of our color matrix there should really be >>> >> >> >> just >>> >> >> >> one function that modifies things based on all those values so as >>> >> >> >> to >>> >> >> >> avoid overwriting what a previous function has done. >>> >> >> >> >>> >> >> >> Still since we have yet to get a working version of saturation i >>> >> >> >> still >>> >> >> >> very much suggest for the time being we revert the current >>> >> >> >> saturation >>> >> >> >> code and just apply a new patch when someone gets an acceptable >>> >> >> >> patch >>> >> >> >> for this. >>> >> >> >> >>> >> >> > >>> >> >> > > >>> >> >> > >>> >> >> >>> >> >> >>> >> > >>> >> > >>> >> > > >>> >> > >>> >> >>> >> >>> > >>> > >>> > > >>> > >>> >>> >> >> >> >> >> > --~--~---------~--~----~------------~-------~--~----~ Lets make microdia webcams plug'n play, (currently plug'n pray) To post to this group, send email to [email protected] Visit us online https://groups.google.com/group/microdia -~----------~----~----~----~------~----~------~--~---
