Sort of like Santa' elves, Sony has something of the sort
situated in an office over in Japan. But these two guys aren't
elves, they're arguably the two people who are the most familiar
with the PlayStation 3 and it's hardware capabilities. Who are
they? Software engineers for SCE Kawanishi and Kanehide — they
seem to go by just those two names.
In an interview with GameWatch, the duo behind the firmware
updates that you download for your PS3, go into detail regarding
their experiences with the PS3 and what's potentially in store
for the console's future firmware updates. In the interview, the
two engineers talk about how they have the ability to improve the
PlayStation 3's backwards compatibility enhancements by another
40%. They also talk about the strength of the Cell and much more.
Gunnerhkjp (of NeoWin forums) was kind enough to translate the
Japanese interview to the best of his abilities. The overall
result of the text is very clear, and not much seems to get lost
in the translation. Feel free to read through the whole
interview, as it is incredibly interesting, but I've also taken
the liberty of bolding the most important portions.
Interview Translated from GameWatch:
The 1.80 software released on May 24th tansformed the PS3 as an
The PS3 gained high quality DVD upconvert using the power of the
Cell processor, DLNA and remote play, and network enhancements.
Even with the past experience of PSPs, not many people expected
the PS3 to change so drastically with a software update.
Here is an interview with the PS3 software engineers, SCE CTO
Kawanishi (GM software platform development) and Kanehide
(Manager in charge of AV development)
I have heard that the upconversion was specifically designed with
anime in mind. Why is that?
The upconversion is not specifically tuned for anime contents. It
works just as well with CG, real life pictures and movies.
The misconception probably came from the fact that we used anime
contents often to check for IP conversion errors.
Upconversion quality does not improve without IP conversion
tuning, which is a lot easier to do with anime then natural
Its more acurate to say that tuning was done extensively with
anime for our convienience, but quality analysis was done with
various kinds of content.
Upconversion works better with content with high SN ratio. Anime
is a good example of such content, but it varies with bitrate,
and also if the content was created digitally or from old film
The quality increase that will come from now on will probably
depend more on the application of noise reduction than the
quality of upconverison.
An example of content with bad SN ratio is TV recording on DVD
recorders. It seems the upconverion doesn't work as well on these
I agree. Quality for normal DVD recorder content with 4Mbps Mpeg2
will depent greatly on the quality of noise reduction.
Ofcourse, it cannot be perfect. The problem is how much noise to
remove. Currently, the NR parameter has 3 levels, so a stronger
NR on such contents may get better results.
The most difficult part of programming the upconverion was the
routine for NR judgement to distiguish between noise and detail.
What is the benefit of NR using the cell processor.
The benefit is that NR can be melded into the entire process of
picture processing, as opposed to it being a separate process
For example, if we judge that a certain part of the picture is
noise or artifact, we can choose not to upconvert that part. With
dedicated chips, this is hard to do.
Also, we have put in the NR at various points for the processing,
instead of applying it all at the end. For example, NR for common
DVD players is applied AFTER the subtitles are inserted. We have
have the NR applied before the subtitle, so that it appears
PS3 also contains many dedicated chips as well, but its
application can be changed with flexibility. Higher picture
quality can be achieved by tweaking the usage.
What improvements can we expect?
I think we are currently making improvements that is easy to see.
I expect much of the upcoming work to be a lot less visible, but
never the less important. Improvement in the NR process is one
A lot of attention is put in the upconversion quality, but as HD
contents become the norm, then picture enhancement for those
contents will become more important.
Maybe the upconversion does not need to be done in real time. How
about high quality upconversion by investing longer processing
That's in the region of re-mastering. Currently, real time
processing is the norm.
It is one of the possibility that we are assesing, but I feel
that the strengh of the Cell processor lies in real time
I guess we could get superior quality by letting the machine work
on it over night. For now, we want to concentrate on real time.
There is a limit to how much can be achieved real time, and
deferred processing is a viable alternative. But such high end
quality enhancements are hard to see by normal people.
But if there is a need, we will do it.
Regarding sound. I think nobody expected the SACD function of the
PS3 to be so complete.
We didn't intend it to be such a featured function, but we
decieded that it would be one of the things we concentrate
initially to get a handle on the Cell processor.
We were only able to do 24bit/88.2kHz degimation, but got it up
to 24bit/176.4kHz. We worked [in] the department which makes Amps
to really get a feel for how sound is processed. When SACD is
playing, cell had nothing much to do. So we concentrated on how
to use the cell to improve the quality of the sound further.
The know how we got from the process for immense, although it
took us away from the the realm of playstation and into the world
of high end audio.
Looking at the SACD playback, it seems emphasis was placed on
sending PCM over HDMI more than the quality of the analog
The reason for that is simple. Although the analog output quality
of the PS3 is very high, it really can't win against AV amps
costing thousands of dollars. Where we can compete in is the
processing power of the Cell. So we concentrated of passing good
PCM data to the amps through HDMI.
PS3 cell processor is a lot better than the chips found in AV
amps, so we figured we will do all the processing on the PS3, and
pass the result to the amps, getting much higher quality.
Ofcourse, the only connection that can pass sound data at that
quality turned out to be HDMI.
How about optical out?
Optical out does not have the copy proctection mechanism
necessary for SACD output. Its unfortunate.
What other functions are you working on? How about virtual
surround for 2ch stereo?
That would be interesting. We are trying to figure out what
For example, what was done before by AV amps can pretty much be
done by the PS3. So functions like virtual surround can be added
to accomodate people who do not own an AV amp.
There is a lot of demand for a jukebox function. How about
functions like lossless transfer into the PS3 and album shuffle?
Also, how do you see the link with portable players?
We haven't really decided on the positioning of the PS3 as a
music player. We don't know how much people want a music player
that requires you to turn on the TV, for example. SACD really
pushed the PS3 in terms of audio, but we have to guage user
But Sony produced many products, and I could forsee the PS3
becoming a "mother ship" to all the equipment.
Ps3 really is a feature rich player, but the basis for all this
is that it is a game console. (laugh)
Q. Why only client for DLNA? I think PS3 should have a server
The primary reason is that PS3 does not have a tuner.
Another reason is that consumers are not yet familiar with DLNA,
and emphasizing the server function may lead to more confusion.
I personally feel that DLNA has a lot of restriction that may
make it hard to use. I think the easiest way is to allow network
browsing of files by using samba, for example.
People who are using the DLNA function right now are people with
high PC literacy. So we tried to make the PS3 DLNA clinet as
compatible as possible with various PC DLNA servers. We also
tested with various NASs with server functionality.
One thing about video file playback. It seems that quality is not
constant, and picutre quality falls from time to time.
We don't have an answer to how much quality we can guarantee from
playback of video files created by PCs. Although there is
standards such as "mpeg-4", there are many cases where
the files contain data that are way off the standard.
We have codecs in place that play back data following proper
standards, and functions to improve the quality of these
picutres. We need more work on handling data that is outside the
For now, out main goal is to allow playback of as many standard
From that point of view, how about playback of WMV and Divx?
The problem with those formats are that they are not industry
standards like "mpeg-4", but format that is locked in
by various corporations.
But I understand the demand (laugh).
How about overcoming the licencing fees for such formats by
selling update for WMV playback at the playstation store for
We don't want to resort to such measures (laugh). And its not a
matter of not implementing because it is by another company.
Your intention for remote play is to make people play outside the
Yes. There are SDKs out to allow game developers to integrate
remote play into their games, so it is a matter of time before
such games arrive.
However, it is now suitable for all types of games. There is a
certain amount of latecy involved.
The output has to be handled by SDKs? How about a simple frame
buffer transfer with encoding to allow any games to be played
To be honest, that's how it works behind the SDK. The reason why
SDK is needed is because the encoding requires processing power
that may or may not be left over from the game. Therefore, we
cannot guarantee that it will work for all games.
On top of that, the controller button configuration is different
How about allowing usage on other than PSP?
Good question. Oficially, no comment. But if you can imagine, it
should be possible on any terminal. The only problem is support
and service quality.
What is the current system software development span?
We try to get one out every quarter.
For how long?
How do you position the PS3 with other home appliances? I feel
sorry for the other AV devices that don't have the benefit of
powerful processor and constant updates. Especially inside Sony,
you have MANY rivals within the same company.
We are in difficult position sometimes, but we don't ristrict
what we do. If we do, we are finished as engineers.
Also, we don't have the benefit of releasing a new model every
2,3 years. So even if we are at an advantage now, they will catch
up with ease if we don't improve.
Ofcourse, we will eventually reach the limit with the current
Cell processor. There is a limit to how much one machine can do.
I hope we can keep on going until we pass the torch to the next
Has anything changed now that Mr. Kutaragi is gone?
The reason I ask is that many people think that good AV centric
updates like the 1.80 would now be over with Mr. Kutaragi gone.
Mr. Kutaragi was a REAL audio visual equipment "geek"
and without him the priority may be different.
I can see why you would ask that question. The timeing (of Mr.
Kutaragi's change and update release) is impeccable (laugh)
It is all a coincidence.
Sony view towards AV technology can never change. It is the
foundation of what we do.
Business choices may change, but the foundation is there.
Which areas do you think the PS3 has the most chance of growth?
For Audio and Visual function, it will evolve in the current
route. PS3 has many input / output capability, and processing the
input for various use will be interesting.
There are other functions for Audio and Visual that are upcoming.
Wait and see.