Hi, Am 20.09.2012 11:08, schrieb Hans de Goede: > Hi, > >>>> Hans, it seems like you didn't pick up these patches up yet... >>>> Is there anything wrong with them ? >>> >>> I've somehow completely missed them. Can you resend the entire set >>> please? >> >> No problem, but I can't do that before weekend (I'm currently not at >> home). >> I've sent these 4 patches on last Sunday (16. Sept) evening. >> Maybe you can pick them up from patchwork ? >> http://patchwork.linuxtv.org/patch/14433/ > > Ah yes, patchwork, that will work. Unfortunately that only solves the > me having missed the patches problem. > > First of all thanks for working on this! I'm afraid you've managed to > find > one of the weak spots in the v4l2 API, namely how we deal with RGB gains.
It seems like this is one of my talents... :( > > Many webcams have RGB gains, but we don't have standard CID-s for these, > instead we've Blue and Red Balance. This has grown historically > because of > the bttv cards which actually have Blue and Red balance controls in > hardware, > rather then the usual RGB gain triplet. Various gspca drivers cope > with this > in different ways. > > If you look at the pac7302 driver before your latest 4 patches it has > a Red and Blue balance control controlling the Red and Blue gain, and a > Whitebalance control, which is not White balance at all, but simply > controls the green gain... Ok, so if I understand you right, red+green+blue balance = white balance. And because we already have defined red, blue and whitebalance controls for historical reasons, we don't need green balance ? Maybe that matches physics, but I don't think it's a sane approach for a user interface... > > And as said other drivers have similar (albeit usually different) hacks. > > At a minimum I would like you to rework your patches to: > 1) Not add the new Green balance, and instead modify the existing > whitebalance > to control the new green gain you've found. Keeping things as broken as > they are, but not worse; and I prefer waiting for the results of the discussion you are proposing further down. > 2) Try to use both the page 0 reg 0x01 - 0x03 and page 0 reg 0xc5 - 0xc7 > at the same time to get a wider ranged control. Maybe 0xc5 - 0xc7 are > simply the most significant bits of a wider ranged gain ? I don't think so. The windows driver does not use them. It even doesn't use the full range of registers 0x01-0x03. Of course, I have expermiented with the full range and it works, but it doesn't make much sense to use it. Experimenting with the device to determine the meaing of unknown registerts, you will notice, that there are several registert which affect RGB. But that doesn't mean that they are suitable for a user control... > Note that if you cannot control them both from a single control in such > a way that you get a smooth control range, then lets just fix > 0xc5 - 0xc7 at a value of 2 for all 3 and be done with it, but at least > we should try :) There is no need to fix registers 0xc5 and 0xc7. The Windows driver sets them to 1, which is exactly the value we are currently using as default value with the blue and red value controls. > > As said the above is the minimum, what I would really like is a > discussion > on linux-media about adding proper RGB gain controls for all the cameras > which have these. > > Note this brings with itself the question should we export such lowlevel > controls at all ? In some drivers the per color gains are simply all > kept at the same level and controlled as part of the master gain control, > giving it a wider and/or more fine grained range, leading to better > autogain > behavior. Given how our sw autowhitebalance works (and that that works > reasonable well), their is not much added value in exporting them > separately, > while they do tend to improve the standard gain control when used as part > of it ... I would say, let the drivers decide how to do things. It also depends on the hardware design. Regards, Frank > > So what we really need is a plan how to deal with these controls, and > then > send an RFC for this to the list. > > Regards, > > Hans -- To unsubscribe from this list: send the line "unsubscribe linux-media" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
