| ↑back Search ←Prev date Next date→ Show only urls | (Click on time to select a line by its url) |
| Who | What | When |
|---|---|---|
| hverkuil | pinchartl: did you accept Helen's patch "media: vimc: fix start stream when link is disabled"?
Not a problem, I'm just wondering who picked it up. Oh, Mauro did. good | [10:13] |
| pinchartl | hverkuil: you're welcome :-) | [10:15] |
| ...... (idle for 25mn) | ||
| hverkuil | pinchartl: do you have a recent image of the omap3 MC topology?
Or at least one that's known to be correct. | [10:40] |
| pinchartl | sure
http://www.ideasonboard.org/media/omap3isp.ps from the Nokia N9 | [10:54] |
| hverkuil | pinchartl: thanks! | [10:58] |
| pinchartl | you're welcome
hverkuil: v4l-utils contains a private copy of the kernel V4L2 headers. so does yavta, and probably other projects. are you aware of any documentation that explains why this is a good practice ? | [10:59] |
| hverkuil | I don't think it has ever been documented. But the main reason is that our API evolves much faster than most subsystems, and distros often are on old(er) kernels and so have older headers.
So apps that try to test for a feature need recent headers and cannot rely on distros. In v4l-utils it is done for the same reason, but there is also is important since it is developed in tandem with the kernel, and compliance tests in particular really need synced headers. | [11:01] |
| pinchartl | I know the reasons, I'm trying to win an argument against someone who thinks that it would be better to point the build system to a location containing the desired kernel headers :-)
so I was wondering if this was documented as a good practice anywhere | [11:03] |
| kbingham | kbingham ducks | [11:04] |
| hverkuil | pinchartl: how do I convert the ps to an image or pdf? Or do you have a link to a ready-made image? | [11:04] |
| kbingham | hverkuil, ghostscript provides a ps2pdf tool ? | [11:05] |
| pinchartl | ps2pdf, from ghostscript ?
:-) | [11:05] |
| hverkuil | ps2pdf cuts off the right half of the image for some reason. | [11:05] |
| kbingham | Doh :( | [11:06] |
| pinchartl | I think you need to specify the bounding box manually
another option is ps2eps and then convert (from imagemagick) or maybe convert handles ps directly ? | [11:06] |
| hverkuil | convert just does the right thing. Nice! | [11:07] |
| pinchartl | are you preparing a presentation ? | [11:08] |
| hverkuil | No, giving suggestions to koike for vimc.
OMAP3 ISP CCDC: is that the debayer block? | [11:09] |
| koike | hverkuil: thanks :) | [11:09] |
| hverkuil | (so long ago, once upon a time I knew that) | [11:09] |
| pinchartl | hverkuil: yes it is
it can also do some other processing | [11:10] |
| hverkuil | CCP2 is like CSI, but a parallel bus? | [11:10] |
| pinchartl | CCP2 is a serial bus
also sometimes called CSI-1 I think | [11:10] |
| hverkuil | Ah.
Thank you | [11:10] |
| pinchartl | https://en.wikipedia.org/wiki/Compact_Camera_Port_2
not sure about CSI-1 sailus: ^^ I think you know more about the CCP2/CSI history than I do | [11:11] |
| sailus | hverkuil, pinchartl: No.
The preview block does de-bayering. | [11:12] |
| pinchartl | sailus: you're correct, my bad
the CCDC handles BT.656, lens shading correction, black level compensation and some form of dead pixel correction | [11:13] |
| sailus | Both CSI-1 and CCP2 are historical. | [11:13] |
| pinchartl | (if I recall correctly) | [11:13] |
| sailus | And compatible, to a certain degree.
AFAIR the CCP2 can use, beyond the regular data/clock signalling, data/strobe signalling that allows a higher data rate apparently. I'm not quite sure which one was first. | [11:13] |
| hverkuil | pinchartl: what did the ISP preview block do? Did it scale down the image? | [11:18] |
| sailus | hverkuil: Debayering. | [11:18] |
| hverkuil | I thought the CCDC did that?
The CCDC feeds the previewer https://hverkuil.home.xs4all.nl/omap3-mc.jpg | [11:18] |
| pinchartl | hverkuil: debayering, more dead pixel correction, gamma/white balance, and there could be something else too | [11:19] |
| sailus | hverkuil: CCDC can operate on bayer and non-bayer data but the non-bayer operation is just pass-through.
More or less, that is. | [11:19] |
| pinchartl | hverkuil: I was wrong, debayering is performed by the preview engine, not the CCDC | [11:19] |
| hverkuil | OK, what's the purpose of the CCDC then? | [11:20] |
| pinchartl | the CCDC handles BT.656, lens shading correction, black level compensation and some form of dead pixel correction | [11:23] |
| hverkuil | It has two outputs, can both outputs be active at the same time?
(it almost certainly is, otherwise the AEWB can't do its work) | [11:24] |
| pinchartl | if I recall correctly they can | [11:27] |
| ........ (idle for 39mn) | ||
| sailus | mchehab: Bom dia! | [12:06] |
| mchehab | moikka | [12:07] |
| sailus | About the imgu driver.
The v8 set is for the staging tree and it seems reasinably ok to me. Do you think it'd be ok still to put it to 4.20, or wait until we have 4.21-rc1? We seem to have 4.20-rc6 now. https://lwn.net/Articles/774420/ | [12:08] |
| mchehab | send PR
I'll decide when I have time to look into it | [12:10] |
| sailus | Ack. | [12:10] |
| mchehab | mchehab is busy today | [12:10] |
| sailus | Obrigado. | [12:10] |
| mchehab | anytime | [12:11] |
| ............................................... (idle for 3h50mn) | ||
| *** | benjiG has left | [16:01] |
| ....................................................................... (idle for 5h50mn) | ||
| mjourdan has left | [21:51] | |
| ↑back Search ←Prev date Next date→ Show only urls | (Click on time to select a line by its url) |
