I am working for 720p and 1080i
using adv212. I am looking for example of parameter setting for this.
for 25Mbps data rate what should be parameter setting of
ADV212 for 720p and 1080i video?
J S Hyanki
The RCVAL for a certain bit rate only depends on the input frame rate--- not the format since you are setting the number of bytes per frame. 25Mbps @ 60hz would be for example:
25000000 / 60 = 416666 bits per frame = 52083 bytes per frame for an RCVAL of 0xCB73 with target size.
Nothing special about the rest of the parameters 720p and 1080i are built-in formats. There are VFORMAT numbers for 720p Luma, 720p Chroma, 1080i Luma, 1080i Chroma in the user/programming guides.
You must use at least 2 chips for HD.
I have seen merging of code stream of Y and Cb/Cr in analog devices ftp site(HDL code).
I want to know that for 1080i application how to seprate Y and cb/cr component
in decoder to feed in both component seprately to both ADV212.
or wether combined stream can be given to ADV212.
J S hyanki
There isn't an application to split them again.
However, you can feed the combined stream to both parts. There is a decoder parameter of size 1 byte at offset 0x4. If it's 0, then the part decodes everything, if it's 1, the part decodes Luma and ignores Chroma, if it's 2, the part ignores decodes Chroma and ignores Luma.
1.Which decoder parameter you are talking about? because in decoder parameter it is not there.
2.Does adv212 upscale the image size even if i feed 1/4 resolution packets.
Just reverse as it does in decoder side by setting DRES register in decode parameter.
1) I'm talking about the ADV212 Decode parameters... the one I'm mentioning is "reserved" so you won't see it in the docs.
2) Nope, it can't upscale.
I got it now. I am trying to pars the code stream before decoding i don't know how much
it will save my bandwidth.
My input video is D1 PAL resolution and I am setting RCVAL in 412B per field for 25fps@8Mbps
and it's working fine.
If I Keep 1/4 resolution of code stream by parsing stream, will size of compressed byte reduce to 13D4 (from 412B to 13D4).
I am trying CIF resolution for bandwidth for lower bit rate.
I want to see how much it can be useful for low bit rate at the cost of resolution.
It's not really linear between the different resolutions. Typically more than half the data is in the LL subband. You'll save data but maybe not as much as you expect at the lower resolutions.
I am trying to see how many byte I can reduce from 8Mbps(412B)
to how much data rate (equivalent byte) ? for CIF resolution.
Has anyone used ADV212 codec for CIF resolution <= 2Mbps.
If yes at what frame rate.
J S Hyanki
I know people have done NTSC/PAL at lower than 1Mbps at full frame rate before. It's all a matter of what quality you are willing to accept. You can arbitrarily set the data rate to whatever you want.
People have used ADV212 for many different smaller resolutions at high compression at various frame rates. Certainly at CIF or lower at 2Mbps or lower-- at frame rates of 60 or lower.
1. Dave If i write 0x01 in 0x04 address in decoder
will it decode LUMA component only(JDATA mode)?
or code stream should be merged from your YCbCr merge block.
2.As you said that people have used ADV212 for lower resolution for lower
bit rate @ full frame. I want to know that whether they have used full
D1(PAL) in input side and used the scalability of code stream or they
have feded different resolution video(CIF,QCIF etc) and did the compression
by using custom mode.
J S Hyanki
1) Yes.. I don't think that it matters if you are using JDATA mode or not but I've always used it in 16-bit HDATA mode.
2) Custom mode compress is more common, but there are customers who have used the scalability and stripped off resolutions before transmission.
I will try for scalability and let you know.
I am trying to use scalability feature of ADV212. for that
i am accessing progression and code style in encoder parameter.
can you please suggest me what should i write in code style register
I want to change the adv212 compressed code stream for CIF resolution.
I want to remove resolution only not quality.
Not sure if I've posted this before but it demonstrates how you can do some codestream scaling using PLT markers (that is what you need to turn on in COD_STYLE for this approach).
I want to know that only PLT bit in CODE style register in parameter should
set or other value like EPH and PPT is require to be change.
if yes what value should be write?
I am writing 0x41 id CODE style.
Only PLT bit is needed. The other markers aren't needed. 0x41 is correct.
Once PLT markers are in there-- you can then parse the codestream on your CPU to remove the J2K packets you don't want.
Thanks for it I will try it and let you know.
I am facing problem in how i will extract resolution,layer and cp from compressed stream.
FF51--------------size i will change as per image size
FF52----------------- i will change transform level
here how i will extract LRCP so that i can remove stream.
I have read literature given by you but i did't get exactly.
You need to use the PLT markers-- that tells you the size of each JPEG2000 packet. You know what order they are in so it tells you where in the codestream the packets you need to remove are. It's covered in the PDF how to do that.
Retrieving data ...