Camera Nitty Gritty - part 9 - What's the deal with video capture on recent devices?
Published by Steve Litchfield at 12:20 GMT, December 9th 2008
Steve Litchfield gets all picky about video capture on Nokia's latest devices. Is there a problem?
If you're into snapping video clips on your smartphone, perhaps of family or friends, for later watching on a big screen, via TV out or DVD, then you'll be interested in the quality produced by the device. Now, we're not talking pro-camcorder quality, but it's got to be good enough to be watchable.
There are two issues here, having experimented with video on two Nokia N96s and an N79. In each case, video clips came out less clear (and with a noisier soundtrack, though that's an article for another day). Incidentally, for these tests I'm going to be using a Nokia N96 with latest v12 firmware, plus my trusty N82, with v30 firmware:
- Rafe's theory is that the latest range of devices (N79, N85 and N96) have cameras set with a video bias towards distant objects. Yes, there's no focussing in any of these phones in video mode, but there's the chance for the manufacturer to preset a specific focus - with sufficient light and depth of field, this can be made to go from a metre or so right back to infinity. In theory. Rafe thinks that Nokia may have set the focus further back on the newer devices for better video quality when shooting more distant subjects.
- In turn, I have a theory that the lesser quality is to do with the lack of graphics acceleration in the newer devices, thinking that this perhaps helps out when capturing video detail in real time.
- Or perhaps Nokia simply haven't got the software right yet!
Can you shed any light on our theories, based on your own tech knowledge and the evidence below? If so, please get in touch in the comments below!
Test 1: Distant subject, good light
Taking video of a house about 60 metres away and then grabbing a frame, cropping and zooming in by 2:1, we get:
Interestingly, the N96 video grab, on the right, is smoother and better coloured than the N82's. I hadn't expected this. Maybe Rafe is right?
Test 2: Close subject (me), good indoor lighting
In this case, holding the camera at arm's length and using the same grab/crop and zoom procedure as above:
Again the N96 frame grab is on the right. This time the N82's video image is much clearer and better 'focussed'.
I was curious as to whether the N82's image was better because of different processing algorithms, so I artificially added 10% contrast and lots of sharpening to the N96's frame grab, in an attempt to simulate what the N96 could do in software if Nokia were so inclined. Here's the result:

The N96's framegrab, even processed, isn't as good as that from the N82.
Focus, acceleration or software?
So, I'm calling out for more data points and theories here. Based on your own video capture experience on other Nseries devices, including the N79, N85 and N96, where do you think the differences lie?
The better quality for distant objects certainly points to Rafe being correct in that the focussing's simply set at a different point. Annoyingly, since you'd have thought that the main reason to have video recording in a phone is to record family and friends - who are likely to be close up. If this is the case, bad call Nokia.
Or perhaps the distant photo test was a red herring and quality's almost always better on the N82 (and N95 etc) because of the extra graphics acceleration hardware helping out?
Or perhaps Nokia simply haven't finished twiddling with the firmware in these newer devices and, once they find the offending bug or setting, all will be right again and results similar to those on older devices?
Your help needed here, I think.
Menneisys and others on the technical bleeding edge, can you add to this debate?
Steve Litchfield, All About Symbian, 9 Dec 2008
Categories: Comment, Hardware
Platforms: S60 3rd Edition
Feature Discussion
Jaggz
Good post.
I'll be honest, video capture on Nseries devices has been effectively in stasis since late 2006. If I didn't know any better (and I don't!), I'd say that it's probably that same old bit of code that was included with the original N93 firmware and the differences are purely down to lens/sensor configurations.
Back in August 2006, If someone had told me that in January 2009 video capture on the very latest Nokia's was about the same, I'd have laughed at their ignorance. Guess I'm the fool for believing that the N93 was the start of something special.
*sigh* I remember a time when Nokia cared more about video capture than 'So Lo'... Looks like those days are gone forever. Time to jump ship perhaps? Does Nokia ever finish anything off properly these days?
James @ Nokia Creative
snoyt
There is so much happening in mp4 compression that it is not really practical to simply compare two random single images side by side. A more in-depth statistical analysis taking the temporal character of mp4 compression would be required. Mpjeg could be compared in such a way.
In a mpeg4 stream there are full jpeg (actually dct I believe) every certain time period, and the time between them is filled with deltaframes using some form of motion compensation for further data reduction. The lossy compression as such is thus very dependent on the prestate of the stream and the input data. Thus when recording exactly the same image, but at different moments, results will vary. Throw in some automagical image contrast, colorbalans and brightness and things get rather fuzzy.
If there is someone working on mp4 compression and has some analysistools to check out the output stream and might even knows how to generate fake input data for the Nokia's. You might get some fancy and reliable numbers.
And my first N95 firmware actually retained it's photofocus from the camera mode onto the video mode. Which was irritating if I forgot. But heaven when I used it properly. I guess that's why they 'fixed' it. The Nokia's are meant for snapshots. And I guess AF does not work well enough for on-the fly video, occasionally it still goes occasionally off skilter for photo's. A visual check seems still neccesary.
Unregistered
"Or perhaps the distant photo test was a red herring and quality's almost always better on the N82 (and N95 etc) because of the extra graphics acceleration hardware helping out?"
Nope. The "extra graphics acceleration" in N82 and N95 is for 3D. It won't help with 2D graphics (video, photographs) at all.
N96 actually has better hardware support for 2D imaging than N95/N82. For video, N96 supports H.264, WMV and MPEG4 on hardware (whereas N95/N82 has hardware acceleration only for mpeg4). N96 also more advanced DSP support. I think your guess about different focusing range is propably correct, hard to say whetever it is intended or 'bug'.
On a more general level, I would say that phone photgraphy has sort of reached it's sweetspot. It's really hard to fit better optics on phones than what N95 and equavalent phones have. Throwing in sensor with more pixels won't help. Software has been pretty much optimized for that level as far possible. Only way forward on software side would mean 'Kodak moments' by 'photoshopping' pictures prettier instead better. Better lenses won't fit into mobile phones unless physical laws governing photons somehow change, and those haven't canged since few milliseconds after the Big Bang. Just compare SLR from 80's and DSLR form '08, and you see what I mean. They might have otherwise very different technology inside, but the optics haven't changed at all.
N93 tried packed some extra room for lenses and zoom with it's strange form factor. I guess problem with robustness of that form, and possibly appeal to customers, made them to abandon the idea.
snoyt
Quote:
Originally Posted by Unregistered
"N96 actually has better hardware support for 2D imaging than N95/N82. For video, N96 supports H.264, WMV and MPEG4 on hardware (whereas N95/N82 has hardware acceleration only for mpeg4).
|
De MP4 is not a symmetrical algorithm. Decompression has very little to do with compression.
Quote:
Originally Posted by Unregistered
On a more general level, I would say that phone photgraphy has sort of reached it's sweetspot. It's really hard to fit better optics on phones than what N95 and equavalent phones have. Throwing in sensor with more pixels won't help. Software has been pretty much optimized for that level as far possible. ... Better lenses won't fit into mobile phones unless physical laws governing photons somehow change, and those haven't canged since few milliseconds after the Big Bang.
|
For snapshots, the Nokia 5Mpixel resolution is pretty balanced for its price/quality. Better lenses and sensors do fit in a mobile phone. Liquid based lenses and other strategy are yet unexplored. Both (lenses and sensors) could improve the imagequality at 5 Mpixels. Optical zoom is unlikely to be introduced unless liquid lenses are used, which is yet unproven technology.
alsiladka
Steve,
In many of the comparisons online, the N85 performed better than the N96. It would have been better had you used the N85 against the N82.
I have no knowledge of the technical details, so i won't comment on the technical aspects of the comparisons, but according to what we have seen in the history of N Series, the quality of the camera's output has improved with the various FW updates offered by Nokia. So i would like to be biased and say that the camera performance would improve over the time.
Kazutoyo
Quote:
Originally Posted by Unregistered
"Or perhaps the distant photo test was a red herring and quality's almost always better on the N82 (and N95 etc) because of the extra graphics acceleration hardware helping out?"
Nope. The "extra graphics acceleration" in N82 and N95 is for 3D. It won't help with 2D graphics (video, photographs) at all.
|
The
OMAP 2420 platform (which N82 & N95 use) have two "graphic acceleration" chips. The PowerVR MBX chip for 3D acceleration and the IVA chip for imaging and video acceleration.
NickAnstee
Steve, have you tried recording the Video to a Micro SD Card instead of the Mass Storage Memory? Nick
slitchfield
Nick, will do. Do you have any observations in this department then? I don't think the issue is one of bitrate, this seems more a 'setting initial focus' bug....
Gerry W
I should confess I switched to an iPhone, but that was more because I wanted to escape 3 and it actually offered an interesting package.
Voice recording, I have about 3 apps which can do it
Radio - well, I love the last fm app
Experience - this is an odd one, with the years of experience nokia has you would expect them to be BETTER in the market than they are...
Camera - honestly, if only it had a flash and video it would be fine as it is ... but yeah, I have a sprog on the way and would think that crap video 'clips' would mean a huge amount to me!
My belief is that the iPhone is the kick that Nokia has needed, but I am not going back to a Nokia device for another 12 months - then I think it might be time to give them a try (or maybe Android).
G
Unregistered
Someone delete the previous post, it was meant for a different thread!!!
Unregistered
There definitly been a change in the size of the lens. I took a look at my n85 to my n95, theres been a slight change in size.
Another factor they have slimmed the newer devices, limiting the space more between the optic lens. rule of thumb, thicker better( i probably learned that here ;) )
My n85, compared to my n95-1 is very poor. in pictures. images do seem poor at a near focual point, but i havent givin it a shot over distance.
11 Comments / Post New Comment