[Mp4-tech][H.264] PSNR is higher than JM??

[Tourapis, Alexis]

Dear Ann,
A common method of comparing two codecs that result in different bitrate
and PSNR values is to plot a Rate Distortion curve. That is, encode your
content at different levels (i.e., using different quantizers), collect
the resulting bitrate (R) and quality (D) for each simulation, and then
plot the curve D(R) . Every different codec would basically give you
different rate distortion curves which you can then compare to find how
these differ. One technique used to compare two RD curves is the one
proposed by G. Bjontegaard, which basically tries to integrate the
difference between two curves that are specified by 4 RD points, and has
been extensively used for standardization purposes. If you are
interested you can check more information here:
http://ftp3.itu.int/av-arch/video-site/0104_Aus/VCEG-M33.doc
With regards to the performance of the JM, note that apart from encoding
decisions (i.e., which modes are used to encode a given block), the way
quantization (and more particular rounding during this process) is
performed can also play a significant role. By default, the JM encoder
tends to quantize coefficients a bit aggressively towards zero, having
in mind a certain distribution, resulting in a lower rate and PSNR but
likely better RD performance vs. quantizing using a rounding offset of
1/2 (i.e., no negative or positive bias). An adaptive technique
(JVT-N011) has been proposed integrated in the software, which may still
lead to lower rate and PSNR values, but would result in even better RD
performance and which you may wish to try out. In any case, I would
strongly suggest that you examine how rounding is performed in your
implementation. Note also that the JM has an option to perform
"thresholding" of coefficients, i.e., if coefficients for a block are
seen as too expensive to code then they are discarded. Although for
intra slices luma blocks are not affected by this, chroma blocks are and
you may wish to disable this feature as well. Just set the parameter
DisableThresholding to 1.
Best regards,
Alexis
  _____  
From: mp4-tech-bounces lists.mpegif.org
[mailto:mp4-tech-bounces lists.mpegif.org] On Behalf Of Ann
Sent: Tuesday, July 18, 2023 3:07 AM
To: mp4-tech lists.mpegif.org
Subject: [Mp4-tech][H.264] PSNR is higher than JM??
Dear experts,
I am doing research on H.264 intra prediction. 
And there is a strange result when doing simulation. 
In the simulation results why the SNR values (Y, U, V) are higher than
JM software.
Although the SNR are higher but the bit rate is more than JM.
This is different with what I think.
If the SNR is better than JM then bit rate should become less compare to
JM.
But why the result is different.
Can anyone correct me what happened here?? Or is there any method to
estimate these 2 values (SNR and bit rate).
Thanks in advance!
Rest regards
Ann
-----------------------------------------
This message (including any attachments) may contain confidential
information intended for a specific individual and purpose.  If you
are not the intended recipient, delete this message.  If you are
not the intended recipient, disclosing, copying, distributing, or
taking any action based on this message is strictly prohibited.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: /pipermail/mp4-tech/attachments/20060718/75da8857/attachment-0001.html