Firmware fixable?

roxics

roxics

Veteran
So are the issues with the D90, the rolling shutter and jell-o effect fixable with a firmware update?
What about Nikon adding a new recording codec like h.264? Is this something that could be added with a firmware update/

I haven't shot with Nikon since the film days so I don't know what their reputation is for fixing camera issues after the release with updates.

Something tells me maybe they pushed this out the door too quickly because they knew the canon was coming and wanted to be the first to announce the worlds first DSLR with video.
 
I don't think that rolling shutter can be corrected with a software update, it's a hardware problem. (I may be wrong)

I'm hoping that they can sort out the poor scaling (Horizontal jaggies) with some kind of software anti-aliasing filtering. For me it's the camera's biggest flaw.
 
Rolling Shutter would most definitely have software (Firmware) limitations.

It's this iteration of the CMOS architecture that is responsible.

I'm not sure about the codec. Depends if they have it hard etched into a dedicated chip or if they have a multipurpose processor.
 
I'm hoping that they can sort out the poor scaling (Horizontal jaggies) with some kind of software anti-aliasing filtering. For me it's the camera's biggest flaw.
Click to expand...

Thats my opinion too. I wish they could do something about it.
 
roxics said:
So are the issues with the D90, the rolling shutter and jell-o effect fixable with a firmware update?
What about Nikon adding a new recording codec like h.264? Is this something that could be added with a firmware update/

I haven't shot with Nikon since the film days so I don't know what their reputation is for fixing camera issues after the release with updates.

Something tells me maybe they pushed this out the door too quickly because they knew the canon was coming and wanted to be the first to announce the worlds first DSLR with video.
Click to expand...


have you tried to encode something with h.264?...ever wondered why image looks muddy?...or if it`s a new codec ... why is not used by industry to render anything..(pro work)...try to submit h.264 to a tv station
 
John,

Going to post the uncompressed grabs shortly, just waiting for the stores to open (8:15am here) to get a longer hdmi cable to get some better light... but I think we are out of luck on the diagonal lines issue. Have a look at "tub" here :

two posts down
 
Last edited:
stoiqa said:
have you tried to encode something with h.264?...ever wondered why image looks muddy?
Click to expand...

That's exactly what I think of the image quality of my HF100 AVCHD camcorder. It looks indeed muddy. I thought it's the tiny sensor and the noise reduction but you say it's the codec itself. Interesting.
 
Test for who has some time on their hands:

Get your 3ccd or HD camera and record 30sec, throw it in your editor of choice and do two renders:h264 and sorenson3(same resolution and same frame rates)

Before you even look at the clips to compare them...look at the size....h.264 should be about 70% smaller. Now get them together and look for yourself . ...let me know what you think about h264 render.(not so happy anymore)

Also ...on D90..there are some good clips and some bad clips....because not everyone knows ho to render properly, or the importance of a final render...so, some of them will look bad.

Quality wise you can work around mjpg with nice results... h264...dont think it`s a good base to start with.
 
Last edited:
H264 mograph quotes:

"
i only skimmed this thread, and i'm not up on all the science behind it, but i'm guessing that h264 was intended for live action / photographic imagery, because 90% of the time i've tested it with more graphic stuff, it noticeably fux with saturation in a very relativist per-shot way (i assume it's due to some form of data-efficient truncating), thus taking a big turd-dump all over your delicately crafted color palettes. if anyone has tips on how to avoid this, i'm all rabbit ears."

"i'm behind sorenson 3 all the way. h264 crushes my saturation values which is unacceptable.

i'm not sure who you plan on sending this reel to but i have yet to work at a shop that didn't have their producers on macs. now i'm sure pc lovin producers are out there but honestly, if you're working in the media industry and haven't installed quicktime on your pc then there is something wrong with you. i know qt 7 had problems on the windows side of things, which is yet another reason to use the solid-as-a-friggin-rock sorenson."

"i second sorenson 3. im not sending anything to clients just to mograph houses looking trying to get jobbies. "
 
Quicktime Player h.264 Bug (http://sprynthesis.com/blog/?page_id=51)

I have an issue here that I’ve been trying to solve for almost a year now. (Any comments can be left on this blog post.)

I’ve been trying to figure out for almost a year now why videos I compress with certain codecs (namely h.264) are playing back in QT with pale unsaturated colors (looks like brightened gamma) while the same videos look correct in the finder preview/FCP/firefox/or the 3rd party “nice player” (when I disable core video in its prefs.)

Take for example this screenshot.

The bottom window is quicktime player playing back one of my h.264 clips, while the top window is the same clip opened up in “nice player” with “use core video” turned off in the prefs. Btw… all videos are compressed with Compressor. I could also open that clip in Final Cut Pro or Firefox and it would look correct (like the top window.)

Here are one and two other examples of other people’s h.264 clips that I have downloaded.

I’m fairly sure that Core Image/Video is the culprit here for several reasons
1) The problem didn’t start until Core Video arrived.
2) The video looks correct in applications that don’t leverage Core Video, or applications such as “Nice Player” which allows you to disable the use of Core Video in its prefs. (http://sprynthesis.com/qt_issue/niceprefs.jpg)
3) The same h.264 clips look fine on every computer I go to that has a video card which doesn’t support Core Video, such as my mac mini (which means there is actually no problem with the compressed clips themselves.)

But I want the clips to look ok on MY computer in regular quicktime, and on my friends computers that have G5’s with graphics cards that support core video.

I also have seen the problem on PC’s as well. The same clip encoded in Sorenson 3 looks great, but the h.264 encoded one looks pale. However, disabling DirectDraw Acceleration and relaunching the clip sometimes makes the clip play normal, but not always.

Originally I thought Apple’s HD trailers didn’t have this problem, on further study they definitely do. They seem to know how to keep the effect of the problem as low as possible though.

So I’m going to venture a guess that the brightness “problem” is going to be showing up with these video cards:

ATI Mobility Radeon 9700
ATI Radeon 9550, 9650, 9600, 9600 XT, 9800 XT, X800 XT, 9850 XT
nVidia GeForce FX Go 5200
nVidia GeForce FX 5200 Ultra
nVidia GeForce 6800 Ultra DDL, 6800 GT DDL
as well as the intel graphics chip in the new mac minis.

Because these are the cards that support Core Video.

Here is a link to a zip file of my short example clip. In this zip file I have the same clip compressed 3 different ways. 1) H.264 2) Sorenson 3 3) Regular Mpeg 4.

The black levels should look about the same on all videos, but the h.264 is much lighter on my screen (see the screenshot of my clip above.)

My workflow
Just in case it matters, I figured I should document my workflow for compressing videos. I’m rendering tiff files out of Maya, putting them together in After Effects 7, and rendering that out to a quicktime movie with lossless animation codec. Then I’m bringing it into compressor and making quicktime movies with h.264 multi-pass.

Is it some sort of color flow issue that I’m missing?

Any help would be greatly appreciated. This bug has been around on Apple’s forums since May with no solution posted.

Lastly here are the specs of my machine

Dual 2.7 G5
ATI Radeon XT 9850
3.5 GB RAM
Dual 23 and 20 Cinema Displays (Aluminum)
Mac OS X 10.4.5

—————
New Research, Big Update

Below is a copy of an e-mail I sent to the quicktime users mailing list, we’ll see what it turns up.

———————-
Hello all,

I’ve been searching high and low to find the solution to this problem, and baffled at how few people seem to be talking about it.

As far as I can tell, any computer which supports Core Video is displaying many of apple’s codecs with incorrect color, including the famous h.264 codec. The problem also appears to extend to some PC’s (I believe Direct Draw Acceleration utilizes the video card to play back quicktime files on some machines.)

You can read about my initial assessment and examples of the problem. http://sprynthesis.com/blog/?page_id=51

In this e-mail I’ll link to further testing I performed over the past 24 hours that clearly identifies my problem.

To follow along with my e-mail, you can download the sample files. http://sprynthesis.com/qtplayer_issue.zip

First here is my workflow.

I am working in Maya (3D app for those who didn’t know) and generating tiff sequences. Those tiff sequences are then put together in a compositing app like Shake or After Effects (I have tested both thoroughly.) I then render those tiff sequences out into a self contained quicktime movie using the (totally lossy) animation codec. No problems through this point. Then I compress the file (either straight through quicktime or using the Compressor application) into an h.264 file. This file then plays back with incorrect colors on my machine, and any other machine that supports core video, but looks fine on those that don’t. The colors aren’t totally wacky, but they are faded and shifted slightly.

This has been brought up in apple forums, and on macintouch and macfixit.

http://discussions.apple.com/thread.jspa?threadID=210793&start=0&tstart=0
http://www.macintouch.com/readerreports/quicktime7/topic2894.html
http://www.macfixit.com/article.php?story=20060331083618226&query=h.264

————

So after reading my web page about this and downloading the sample files, you can test this for yourself here by breaking out the DigitalColor Meter application from your utilities folder (set the pop-up menu to “RGB As Actual Value 8-bit).

1. Inside my folder of sample files, open the first movie called “1.single_tiff.mov”. This is simply one of the tiff files saved as a self contained movie in quicktime.
Mousing over the squares (http://www.sprynthesis.com/qt_issue/likethis.jpg) will give you your base values which all the other files SHOULD match somewhat closely. Use this as a reference file to compare all the other quicktime files to.

The bottom row should be various levels of greys, the next row up should only contain blue values, and so on with the red and green rows. The rows above that are various combinations of RGB values just for further testing.

————

2. Now move on to the “2.ae_animation.mov” file. This is the tiff sequence compressed out of after effects using the animation codec. The values of these squares should be identical to that of the first file, and they are.

————

3. Now look at the “3.h.264.mov” file. The results you will get for the RGB values of these squares will depend on whether or not your computer supports core video. If it doesn’t support core video (like the PowerPC mac mini) then your RGB values will be extremely close to that of the source file. (Hooray! The problem doesn’t affect you.)

If your computer does support Core Video, then you will see that all the black levels - except for the pure black and white (its a gamma curve problem?) - are brighter than they should be. When you mouse over the squares that should be solid RGB values, you see anywhere from trace amounts to extreme levels of other colors showing up. This is quite a problem! What’s worse? Calibrate your monitor instead to a gamma of 2.2 (like most pc users and a lot of pro mac users do) and the color shift is much worse.

Now is the fun part. Open that same h.264 file with Firefox, or download an application like the “Nice Player” (and disable core video in the Nice Player prefs.) The color problems dissappear and the file looks like its supposed to (like those without core video, the rgb values don’t match exactly, but at most they are off by a hair… very acceptable)!! Well crap… what’s a guy to do? Use Sorenson 3?

————

4. Open the “4.mpeg-4.mp4″ file. The RGB values should look pretty good in quicktime no matter what computer you view them on. Although when viewed in Nice Player or Firefox the RGB values actually seem a little low on the color squares. (Maybe this is because no ICC profile is embedded in mp4 files?)

———–

So you should be able to see my problem pretty clearly as long as your computer supports core video. My question is what is the solution? Whoever encodes the Apple HD trailers seems to know. Those videos look great on computers with and without core video.

Anyway, that’s my 2¢. I am determined to understand what’s going on with my color conversions. Anyone who has any ideas for a solution, drop me a line! I really want to be able to trustfully use h.264 instead of falling back to Sorenson 3.

Thanks,
Robert Spryn
——————–
 
I'm sorry, but I just have to say something about H.264 and the trashing that it's gotten from a certain poster here.

Basically, you're wrong. You're very, very, VERY wrong.

1) H.264 is an industry standard designed to replace the aging MPEG-2. TV stations might not accept H.264-encoded video, but that's because a) they're slow to catch up with new technology, and b) MPEG-2 was established years ago as the standard for DTV/HDTV. MPEG-2 has proven sufficient for ATSC broadcasts, so the industry is going to follow the path of least resistance and avoid switching over to H.264 (or any form of MPEG-4) wholesale in the near future.

2) Blu-ray releases are commonly encoded with H.264, proving concretely that it is an industry standard delivery codec, not something subpar and unprofessional.

3) Sorenson 3 is actually an implementation of H.264. The story behind this is pretty convoluted, so I won't retell it here, but you can read all about it on this page: http://www.drunkenblog.com/drunkenblog-archives/000312.html

4) To be more precise, H.264 is a standard for codecs, not a particular codec in itself. There are many implementations of H.264, just like there are many MPEG-2 codecs. Some implementations are better/worse than others. Apple's implementation is widely regarded to be disappointing. Thus, the fact that Sorenson 3 consistently produces better output is hardly surprising.

5) H.264, like other MPEG standards, is meant for delivery, not acquisition. In other words, it was specifically designed and optimized for encoding finished video projects for distribution rather than for camcorders. This is why video from AVCHD camcorders have serious drawbacks, just like with HDV (which is MPEG-2).

6) The Canon EOS 5D Mark II records video with H.264.
 
Last edited:
Who am I to fight the evolution wave...I was just posting samples from industry people that actually tried to encode H264...and based on my render experiments.

5 and 6 points they go head to head.

have you tried encoding h264 and sorenson 3...let me know how it looks.I`m no expert,that`s why I quote people smarter than me...and.. I dont like the final render with h264.


"How does it compare to the current crop of H.264 encoders?
*


- featurewise its clearly A LOT worse than most avc codecs, like x264 or nerodigital, which offer a much more complete implementation of the mainprofile (as i said: apples codec is more baseline profile + b-frames, as a main profile encoder...)
- also if basic things like cropping dont work than this means that you will loose quality you could easily save
- also a two pass encoder has to be able to hit the target bitrate on the spot
- i didnt really made a frame-to-frame comparison with other codecs, but the quality of the 900kbps encode didnt positively surprise me at all. at such a relatively high bitrate i would have expected a clearly good result and not a pricture that tends to blur on some parts. i wouldnt be surprised if x264 or nero provide better quality at the same bitrate, especially when you enable more advanced features like cabac, which are said to bring the bitrate down by about additional 10% losslessly (yeah losslessly)
i also made a quick compare with the 600kbps sample with nerodigital and apples codec tends to be clearly worse after a quick view

this whole sitaution reminds me that apple also only provided a VERY basic mpeg-4 part 2 encoder, only supporting simple profile...
i can only hope that apple will continue to work on their avc encoder and not leave it at it is now (for the marketing machine there is enough already it seems: vbr, b-frames...), so lets hope that their avc decoder at least supports the full main profile (cabac, loop, weighted pred aso...), cause if not i would say qt7 will be as useable for avc as it has been for mpeg-4 part2: not really useable laugh.gif"
 
...actually on point 4, you agree with what I`m saying

My question is....where is that better h264 implementation that is better than QT?...(not on Vegas, AE or PPro)
 
evilfoxhound said:
I don't think that rolling shutter can be corrected with a software update, it's a hardware problem. (I may be wrong)

I'm hoping that they can sort out the poor scaling (Horizontal jaggies) with some kind of software anti-aliasing filtering. For me it's the camera's biggest flaw.
Click to expand...

yeah the rolling about hardware. the cmos isn't scanned fast enough. that's why i think that video on a d300-like camera will have less rolling.

but the aliasing should be fixed. they could do another algorithm for that.

there's no way to get h264 thru firmware. all cameras use hardware codecs for that, not the general purpose cpu.
they could compress less the mjpeg i think, as an option.
 
stoiqa said:
...actually on point 4, you agree with what I`m saying

My question is....where is that better h264 implementation that is better than QT?...(not on Vegas, AE or PPro)
Click to expand...

Isn't this that old chestnut that I found while trying to output a decent blueray?

There is not on mac?!

I output a superb HD-DVD using dvdstudio pro on a dvd-5, and for the life of me tried every program qt/compresser/toast/premier etc to get a decent h.264 blueray burn, and couldn't find anything that compared? Scratching my head as to how hollywood is producing them then, I spent a week of testing and research only to find

- there is no commercial-grade h.264 encoder for mac. Yet.

Some people were claiming premier was the closest.

Hd-dvd was a culmination of years of mpeg development, so it was not much of a stretch to put a commerical grade HD encoder in dvd studio pro. When HD-DVD died, apple was forced to start fresh with blueray, and has just sighed for the meantime, letting premier and toast pickup the slack. Compressor just does not put out a high qual h.264 to blueray specs. It puts it out, but.... Even trying to find something that would then burn that output without trying to re compress it was fun and games.

Leaving mac users with all the tools except to compress a decent blueray...

The web was full of people going WTF! Toast might cut it for your home videos, but not to hand to a client.

I ended up taking my output to bootcamp and winxp/vista using DVDitPro for h.264 compression. That was the only product I could find to do a decent job. It uses a same/similar h.264 compression engine to Scenarist BD which is $40k!!!!

It seemed ridiculous to me that I could produce a quality HD compression in DVD Studio pro for next to nothing, but it took $40k to produce the same thing in blueray, and DVDitPro was the only thing that came close, and it meant taking it to windows.

But after a week of reasearch and testing, that was the conclusion.

So if the jist of some of the posts above is that quicktime h.264 is crap. Yes. Is there anything else? Yes, one cheaper product on pc, or $40k scenarist.

Will there be something on mac? Eventually... The research I did is two months old now, but it didn't seem like anything quick was coming...

But it doesn't mean h.264 is necessarily **** on the whole. Hollywood movies on my ps3 on blueray look great.

H.264 for recording... well... I have no information there. Could be good/bad depending on what's implemented I guess... Are their precedents of good/bad, or only bad.. ? :)
 
H.264 is awesome and a huge leap forward in video encoding standards. But you have to remember that it's just a standard, not a codec in itself. Apple's implementation may reflect poorly on H.264, but the fact that the output is not very good is not the fault of the H.264 standard but rather the fault of Apple.

The best easily-obtainable H.264 codec is x264, which produces the highest quality video at among the lowest of encoding times. The Windows version is extremely easy to find, but there's also a QuickTime version for Mac users that can be found here: http://www.macupdate.com/info.php/id/20273

This is the solution to your encoding woes with H.264.

Just because H.264 is not an ideal acquisition codec does not mean that an H.264 camcorder can't capture great images. If properly implemented (which, sadly, is not the case for virtually all AVCHD camcorders), H.264 would trounce HDV at the same bitrate. The video so far released from the Canon EOS 5D Mark II, which records H.264 at 40 Mbps, speaks for itself, outclassing the video from camcorders five times its price (though still falling short of the RED One).

In other words, if Nikon had employed H.264 compression on the D90, and implemented it with at least the skill of Canon, the video quality would be superior to MJPEG.
 
Last edited:
You must log in or register to reply here.
Back
Top