Ive found a program called VLC http://www.videolan.org/vlc/ that can show the stream (full screen) from the 313 & record it to disk. Im not sure if it can handle 90% yet but so far it looks very good. no skipping & no Fat32 file size limit! It works also on Linux but Ive not tried it yet & can also work with Ogg so it should be fine with the 333.
There has to be some problem (there is always some problem) I will now go & test it a bit.
Yes the first one is I would love a control panel for the camera that works in windows ....if this is the solution.
there is a kind of grid pattern with the picture although it seems very sharp.
I have made a whole string of files that can not be opened. Im guessing you must record raw & then give the file a name but I never seem to get a file if I select raw.
You can change the cache size so there is a longer or shorter delay.
To stream just select the network tab , enter the rtsp address & then check the steam output box.
After watching the video with the grid (is this something to do with the Bayer filter?) I see the picture is much sharper so I think this is not a problem if the files render ok with the grid.
The files that VLC produces seem only to be usable in the VLC program & Mplayer. The sadly is no avi option for the output file. The videos Ive made today seem perfectly smooth & the camera seemed happy to push out 85% quality, full screen although it started to slow up with 90%.
I will see if I can get Mplayer to convert the file into frames
I hope I'll get a little time to make a program that will build ".mov" headers for the captured files - I'll reuse that code that is in the camera now (for video clips - I made it 2 years ago).
As for the "grid" - can you show an example frame? The color conversion from the Bayer pattern is not really good in the 313 - for interpolation it uses just the pixel and its 8 neighbours. In the 333 (where FPGA is much bigger) it will be 5x5 core, not 3x3 and the external memory accesses are designed for this. But in this early code we have now it is still the same as in 313 - outer pixels are ignored.
I'm converting the file to jpeg frames with Mplayer so I can actually try rendering it. Ive never seen an image in video as clear as this but when you look close it looks like interlaced TV on a PC!
The VLC player seems like a great alternative to Mplayer but just lacks the ability to make a usable file from the Jpeg stream!
I think this is a true eureka moment! The grid seems to fade away a little when rendered to Xvid & the extra resolution helps make the finished file seem extremely detailed or clearer (from a distance). With VLC I can use my external NTFS drives & not have to think about a file limit. I can then fire up Mplayer to convert the file to Jpegs & then bring this all into Vegas to edit into a nice Xvid avi. The frame rate seems to stay stable also (although I need to test this more as 90% did effect it).
I do not understand how the image can be so different in windows with this player? Its different but I like it & I think our customers will also!
Is the "original" frame received directly from the camera as static JPEG image or is it made from the stream after some processing?
If it _is camera static image_ it looks like very poor compression quality, or - even errors in the compressor (if clock frequency is too high for it) - try reducing fclk and see if the problem goes away.
It really seems that FPGA compressor clock is too high for this camera - try lower values. And you do not need to change sensor clock - they are independent of each other.
I tried lower values & it made no difference. It seems VLC does something to the signal from the 313L? If I use the live CD there is no grid & there is no grid if looking at the image (screen capture) in windows from within the 313 fpga.
In that case - it can be VLC - I never tried it myself. Just the images you linked seemed like there was some problem in DCT/quantization I've seen with the FPGA with too high clock.
Other possible explanation (if it appears with some players, but not with others) - difference in quantization tables between the compressor and the decompressor. Why it could happen - I have no idea and I've never seen it so far with any of the decoders we used.
I think Ive figured out how to get a high steady fps output (at least inside!). You must open the iris to its max & use a fast shutter speed. The slower the shutter the slower the fps go. Ive been filming outside mostly with the Iris closed half way. I probably had the shutter speed to slow to compensate. I Will try tomorrow out side & keep my fingers crossed !
Exposure can influence frame rate only if it is longer then original frame period (1000ms/fps) - you can see the actual frame rate that shows on the "?" page.
If it is not the case, then exposure can influence the actual frame rate only indirectly - i.e. with higher exposure you get more detailes, frame size increases, streamer/network/computer limits the bandwidth and some frames are dropped.
Log in to post a comment.