D-Movie = D-Movie = D-Movie

[Martin Koch]

There's always some uncertainity and wishful thinking every time a new Nikon model with video comes out. Although I see that the data rate hasn't changed a bit I think that maybe somehow Nikon did a miracle in the new model. Wouldn't it be good to get hard facts?

Well I found JPEGsnoop (http://www.impulseadventure.com/photo/jpeg-snoop-uses.html) which allows to get all the fine details from JPEG photos or MJPEG video files e.g. you can see the actual Luminance and Chrominance quantisation tables that determine JPEG quality. It's like watching CSI.

I just compared native .avi files from the D90, D300s and D3s. There's not the slightest difference in the quantisation tables of the three cameras. They may have improved the downsampling on the newer cameras but the MJPEG compression quality is identical in all Nikon models.

This is nothing new but I thought it's interesting to see a technical proof.

[Tracey Lee]

Hmm maybe I should get a D5000 then and sell my D300s just to make a few bucks for gear. Does anyone know if the D5000 skips lines like the D90 does?

[Martin Koch]

Hmm maybe I should get a D5000 then and sell my D300s just to make a few bucks for gear. Does anyone know if the D5000 skips lines like the D90 does?

The D5000 is the same as the D90 in regard of movie quality. The D300s has a better downsampling (no stair steps) but you can bust the stair steps of the D5000 using the famous Lee Wilson Fix. So it indeed does have little advantage owning the D300s if all you want is filming. Other than obvious built quality, better stills and so on. This is weird but when it comes down to pure D-Movie quality the D300s has no significant advantage and the D3s has only the additional advantage of a (hidden) manual video mode.

Other things that speak for the D300s are the external mic input (although it is sampled at shockingly low 11 kHz) and the arperture mode.

[Tracey Lee]

Martin,

Do you know if the exposure lock on the D5000 stays locked even after you stop recording?

[Martin Koch]

Sorry Tracey, I sold my D90 a year ago to make a few bucks for gear and I just can't remember if I had to lock the exposure before each and every recording or only after re-entering LiveView. Anyway do yourself a favor and test a D5000 at a camera shop before you make such a decision.

[Zim]

Wait a couple of weeks. Nikon is about to announce some new cameras.

Hmm maybe I should get a D5000 then and sell my D300s just to make a few bucks for gear. Does anyone know if the D5000 skips lines like the D90 does?

[Ruzo]

Hmm maybe I should get a D5000 then and sell my D300s just to make a few bucks for gear. Does anyone know if the D5000 skips lines like the D90 does?

I would advise you to personally compare them first.

[mhill310]

I have a d5000 and it is the same as the d90, you can lock exposure for as long as you want as long as you dont leave live view. It takes the same video. I like it alot.

[Jim Klatt]

Wait a couple of weeks. Nikon is about to announce some new cameras.

According to the Nikon rep.'s interview on NR, it will only be Coolpix cameras that are announced in Feb.

[damonb]

My understanding is that you can lock, go into live view, go out and in again and exposure will still be locked. You have to either unlock it, or turn the camera off and on to escape.

[Sean Michael]

I just ordered a Nikon factory refurb D5000 from Adorama (with 18-55mm VR kit lens, free shipping, & no tax) for $524.

I was ready to buy a D300S but couldn't justify the extra $1000. Same sensor and image processor. Same 720P 24fps. As the thread title aptly states, D-Movie = D-Movie.

Although it's the cheapest selection, the D5000 actually has one advantage over the others for video -- the articulating screen, which should prove handy for low angle shots.

[Zim]

what I read over there was Feb 3 Coolpix announcement then the week after more stuff. But that is only a rumor too!!



According to the Nikon rep.'s interview on NR, it will only be Coolpix cameras that are announced in Feb.

[Lammy]

Guys. The D90 and D5000 is super mushy and looks like SD compared to the D300S and D3S.

The D90 and D5000, If you have fully automatic lenses like the kit lens, you're limited to shooting D-Movie Mode at wide open aperture.

On a side note, I still love my D90 to bits =]

[Martin Koch]

Guys. The D90 and D5000 is super mushy and looks like SD compared to the D300S and D3S.


But how can this be? Are there other secret ingredients in MJPEG compression? The quantisation tables, chroma subsampling and data rate are identical. The source image they start with is better but thats the only thing I can think of.

You seem to have both cameras. Could you please post two frame grabs from the D90 or D5000 and D300s equiped with the same lens of the same well lit scene, shot at the same time? This would be very helpful because seeing is believing! I will apply the Lee Wilson Fix to the frame grab of the D90 and repost it.

[PhantomVideo]

The D5000's screen is half the res of the D90's which makes cirtical focus, pretty dam hard

[Lammy]

But how can this be? Are there other secret ingredients in MJPEG compression? The quantisation tables, chroma subsampling and data rate are identical. The source image they start with is better but thats the only thing I can think of.

You seem to have both cameras. Could you please post two frame grabs from the D90 or D5000 and D300s equiped with the same lens of the same well lit scene, shot at the same time? This would be very helpful because seeing is believing! I will apply the Lee Wilson Fix to the frame grab of the D90 and repost it.


I've only got the D90 and 7D right here, the D3S is in the studio 60 miles away and I'm too lazy to get access to it.

But I assure you there are huge differences. It's nothing to do with compression. It's the process before that.

Probably the way processor and the debayer algorithm is set. The D90/D5000's is like on a quarter debayer, not to mention some very strong Optical Low Pass Filter going on in the works.

[Martin Koch]

I've only got the D90 and 7D right here, the D3S is in the studio 60 miles away and I'm too lazy to get access to it.

But I assure you there are huge differences. It's nothing to do with compression. It's the process before that.

Probably the way processor and the debayer algorithm is set. The D90/D5000's is like on a quarter debayer, not to mention some very strong Optical Low Pass Filter going on in the works.

Thanks for your experience with the D3s. I believe you. :beer:
I still wished I could see it myself especially a comparison between the D300s and the D90.
No one has ever shown that.
Or the D300s compared to the D3s. I only assume its the same quality but who knows.

By the way I also looked at the quantisation tables of Panasonic GH1 or Pentax K-7 MJPEG files. They show much better compression quality (although only 4:2:0) like 75% (Nikon) to 90%.
I wished the D300s would have these tables built in.

[Tracey Lee]

I own the D300s and I can say that it is pretty easy to focus using the LCD. I am thinking also that not having to use the Lee Wilson workaround and also have a SD and CF card slot is worth the extra money. Post production is fast and I don't need to trade out cards since I have both a 16gb SD and 16gb CF card in there.

[bang]

The D5000's screen is half the res of the D90's which makes cirtical focus, pretty dam hard

agreed, articulated screen is nice feature, but its really hard to see whats in focus

[killacam]

isn't the jello on the D300s and D3s supposed to be improved a lot over the D90 and D5000? also, when you say "the D3s has only the additional advantage of a (hidden) manual video mode" you also have to mention the fact that it is full frame and not dx (along with the low light improvement).

[Martin Koch]

you also have to mention the fact that it is full frame and not dx (along with the low light improvement).

You're right I didn't think of that.

[stip]

The D5000's screen is half the res of the D90's which makes cirtical focus, pretty dam hard

I have a Pentax K-x which has a 230.000 pixel LCD, so not even 1/3 of D90 but strangely I can focus with it quite as good (or as bad :happy: ) as with D90.



isn't the jello on the D300s and D3s supposed to be improved a lot over the D90 and D5000?


http://vimeo.com/8009537

I don't even see a difference (jello-wise) to the RED in this comparison. That's unthinkable with the D90...
(http://vimeo.com/8009537)

[mattsand]

The source image they start with is better but thats the only thing I can think of
um, that's the *only* thing you can think of? so nothing significant then? who cares about the source image anyway. ;-)

[Sean Michael]

How much better can the source image of the D300S be over the D90 / D5000? They all use the exact same DX sensor, same Exspeed processor, same active dynamic range. I've looked at still images taken from each of these cameras (on the camera review sites) and they are pretty much identical. The differences between these DX models seem to be features (control buttons), build quality, LCD screen, and in the case of the D300S a mic input -- but not image quality.

The D3s is a full frame camera so it's a different beast. The source image should be better. Of course, it also costs about $4500 more than a D5000.

[Martin Koch]

um, that's the *only* thing you can think of? so nothing significant then? who cares about the source image anyway. ;-)

:Drogar-Happy(DBG): When I wrote this I *only* thought of the better downsampling and not as BandanaDan says better overall source quality.


How much better can the source image of the D300S be over the D90 / D5000?

You're correct but in the case of D-Movie the source is the Live View feed and it seems this signal is of exceptionally bad quality on the D90/D5000.

[mattsand]

How much better can the source image of the D300S be over the D90 / D5000?
i think you're too narrow in your definition of source image and what's involved in that. the debayer, noise reduction and scaling are all extremely important, and that's where the d90 sucks. just zoom into the live view and you'll see that it's already crappy before the jpeg compression. or just do the test that was suggested after somebody first used jpeg snoop on a clip and found out the "quality setting"; resize and compress one of those stills to 1280x720 at 75% quality and that's how good the image can and is supposed to look.

[mattsand]

example. one d90 still resized and compressed to 65kb (i didn't have percentage in the software so i divided the bitrate of the video by 24 and made it that size) and one frame grab from video (less compressed to preserve the original quality). there's quite a difference and to me there's no doubt it has very little to do with the jpeg compression. make sure you view them at full size, even though the difference is fairly obvious even scaled down.

[hepabst]

So what did Nikon do differently with the D300s? Do any of you folks have a D90 and a D300s and understand the differences in image quality?

[Martin Koch]

example.

Yes the difference is pretty obvious and I've posted that long ago (http://www.dvxuser.com/V6/showthread.php?t=148666) (my object wasn't that cute though).

I thought for a moment you're showing a D300s and D90 frame grab.
I wished I could see such a comparison between a D300s and a D90 because from what I've seen from Vimeo downloads I doubt that the D300s grab will look this good. Too bad you don't have a D300s.

[Martin Koch]

... just do the test that was suggested after somebody first used jpeg snoop on a clip and found out the "quality setting";

This has already been posted by someone here? Sorry I missed that.

[mattsand]

i know you did, what i'm trying to point out here is not that the quality sucks, but that it sucks much more than the jpeg compression could possibly explain, which is what this thread is about. the still frame is recompressed at the same quality setting and resulting size in kb as a video frame. this is how good the video image can theoretically look, since the compression is the same and i'd say a properly downsampled 12mp image is pretty much ideal source material, and never better since the jpeg compression becomes the limit.

(i know, that leica is cute. i bought it for her) :-)

[mattsand]

This has already been posted by someone here? Sorry I missed that.
not the d3x but i did for the d300s and of i'm sure someone did for the d90, maybe even yourself? when that happened several people did that very experiment, some showed that even cropping the image to 1280x720 before compressing resulted in a better image, suggesting that it's not just bad scaling but some other cut corners as well. the faster the processor the less cutting of corners necessary, so it's not surprising at all that newer cameras perform better despite similar specs.

[Ruzo]

Yes the difference is pretty obvious and I've posted that long ago (http://www.dvxuser.com/V6/showthread.php?t=148666) (my object wasn't that cute though).

I thought for a moment you're showing a D300s and D90 frame grab.
I wished I could see such a comparison between a D300s and a D90 because from what I've seen from Vimeo downloads I doubt that the D300s grab will look this good. Too bad you don't have a D300s.


I don't have my D90 anymore but did put a couple of comparisons a long time ago, when most people did not care. I had just received the camera and some settings might not have matched but you can clearly see the "mushiness" or "melting paint syndrome" of the D90 compared to the D300s.

http://www.dvxuser.com/V6/showthread.php?t=184111

[Martin Koch]

i this is how good the video image can theoretically look, since the compression is the same
(i know, that leica is cute. i bought it for her) :-)

I completely understand what you show us here but in the case of the D300s the quality still doesn't seem to be there. I think dpreview.com brings out the weakness of a camera quite good by filming brick buildings and detailed traffic scenes. Take a look at the very first D300s .AVI (http://www.dpreview.com/reviews/nikond300s/page19.asp).? Although better than the D90 .AVI clip (http://www.dpreview.com/reviews/nikond90/page18.asp) (the first one again) it's not quite there isn't it.


Sure I referred to the camera :-) In fact I didn't even notice it!

[Martin Koch]

I don't have my D90 anymore but did put a couple of comparisons a long time ago, when most people did not care.

I seem to have missed a lot of good posts! This is a great comparison and a first time for me to see the same scene from both. Thanks a lot.
My conclusion is that the D300s or D3s movie quality has improved but not so much that it makes me want those models. I better wait for the next (or a later) generation.


@mattsand: No it wasn't me I discovered JPEG Snoop just a few days ago and I see in this thread that you already talked about the quantisation tables back in September 2009. :-)

[mattsand]

i actually talked about them in sep 2008, but whatever. :-) i'm not trying to prove that the d300s is great, it's actually not, just that it's entirely possible to get a good image even with mjpeg compression. that's not the bottleneck here.

[Tracey Lee]

I played around with the D5000 and it's not too hard to get a good focus and image. Then with Mattias's plug in the image is about the same as the D300s. So I decided to sell the D300s and get some more sound equipment to go with my D5000.