The end of 35mm adapers and hv20s....

you probably said so but others seem not to believe... i was just posting links. trust me, no matter what i say i'm one of your fans here.

do a test with a custom curve. if it works as with photos you'll be very pleased.
compress your highlights a bit, up the midtones. stuff like that

here's a link for photo buyers
http://fotogenetic.dearingfilm.com/custom_tone_curves.html

in fact get the demo from nikon's software called silver efex pro. it has presets for lots and lots of b&w film types. then get the demo from color efex, it has color film presets as well. copy the exposure curves from there.
since i shoot photos raw picture controls have very little use for me. but this video feature changes the whole idea. picture controls become a must have.
 
Last edited:
Steve Castle said:
This is under the assumption that rolling-shutter compensation will be applied without prejudice on every scene; which as you say can cause problems. However, if a reference is provided by an accelerometer or other device, then the reference can be used by the rolling shutter compensation algorithm to compensate relative to movement. So the algorithm can take into consideration if the camera is panning left/right/up/down, still, etc. There are also several software-based solutions then the simple one provided in the paper in question, like other combination spline/block based motion estimation methods, 3-D feature correspondences, etc; which can better compensate for non-static objects, but are computationally more complex.

Of course nothing is that simple, there are several technical and cost considerations as well. If it could be applied easily, it would have been applied already. But a lot of companies are heavily invested to make sure that CMOS is the future imaging sensor. CMOS deficiencies like modulated light effects and skew created by rolling-shutter are being addressed, there are both hardware and software solutions to all of this, the real question is which one can be most effective as well as cost-efficient.
Click to expand...


I wasn't considering every scene, just scenes with distortion. And sadly none of methods mentioned is even near adequate.

Even for the sake of arugument, you get perfect "corrections" you will still have missing information which you can not re-create.

Only one method is known to work, its called global shutter :D

Now believe me, I'm looking for a solution too, I just don't think there are currently any good ones; from an engineering POV, its the source where the problem lies not its after effects.

EDIT::

3d feature correspondences, I can't even see how that would be possible, even the sparse points recreated from a single moving highly calibrated camera is limited, calculating features points from a distorted image would be near impossible, unless you are talking about a multi camera set up, thats different, but then you would still have to solve the orrespondence problem.
 
Last edited:
ando said:
http://www.dpreview.com/previews/nikond90/page2.asp

• Exposure: Determined with matrix metering utilizing output from the image sensor
• Exposure lock available
• Exposure compensation available in P, S, A, M modes

You can also apply Picture Control settings to movies (sharpening, custom jpeg exposure curves, saturation, hue) or one of the presets
si basically you can adjust all image parameters (custom curves are raw->jpeg mapping curves) and also use +- compensation with video.
Click to expand...

Thanks, but I guess I am still missing something. If for example I put the 50mm 1.8 lens on it with the aperture ring set on auto, what f-stop will the camera set it to? And what about if I change the f-stop manually, will that work? And I guess the same would apply to any of the older manual lenses?
 
Matthew Bennett said:
I read its possible to shoot at almost 23fps (with the 'motor drive' or continuous mode) with this camera using the small jpeg, it would destroy the shutter in no time, but at least you would have a decently sharp looking image, no? and almost 24p

Anyone tried using the motor-drive at max capacity? I often blast a d70 at nearly 12 fps. Of course I broke the shutter after 30,000 pulls or so...
Click to expand...


The D90 is tested to 100,000 shutter actuations.

Shooting at 23fps would take the D90 past it's life span in one hour and 15 minutes !
 
Erik Olson said:
I haven't read through all the pages, but this is a cool new tool. Can anyone tell me why I'm seeing what I think is a rolling shutter issue in the trains clip? Again, sorry if this was covered on page 186 or something, but I simply can't find it. I guess I'm wondering how a locked-off shot seems to be wavering.

e
Click to expand...


Vibration from the trains.
 
zcream said:
Hi Lee. It could be just the problem with the compression. We are trying to fit 1MB i.e. 1280x720 @ 24P inside 12.5 mbps while DV fits 500K SD at 25 mbps with a more advanced JPEG compression scheme..
Standard DV is not only 4x times less compressed but also a better compression..

If only we can get the 4:2:2 or 4:4:4 HDMI signal..
Click to expand...

Good point.

I suspect besides the (poor) choice of Motion Jepg - the way the image is extracted from the sensor plays a part.

Binning + old and inefficient codec on a low quality setting = poor image.
 
Anhar Miah said:
I wasn't considering every scene, just scenes with distortion. And sadly none of methods mentioned is even near adequate.

Even for the sake of arugument, you get perfect "corrections" you will still have missing information which you can not re-create.

Only one method is known to work, its called global shutter :D

Now believe me, I'm looking for a solution too, I just don't think there are currently any good ones; from an engineering POV, its the source where the problem lies not its after effects.

EDIT::

3d feature correspondences, I can't even see how that would be possible, even the sparse points recreated from a single moving highly calibrated camera is limited, calculating features points from a distorted image would be near impossible, unless you are talking about a multi camera set up, thats different, but then you would still have to solve the orrespondence problem.
Click to expand...

Feature correspondence analysis I referred to is tracking software to find the point correspondences in the sequence that can be used to identify objects in the scene. If the background is static and can be easily identified and subtracted and shouldn't be an issue. Many of these techniques can be used in conjunction with each other, and feature correspondence was actually used in earlier research to the common/faster method today that uses analysis of distortions in interframes/intraframes. Also, most rolling shutter algorithms do both an analysis of global motion estimation as well as identify local motion (such as that of an object moving across the screen).

Hope that helps, I mentioned it because I am familiar with certain groups that are working on this currently. There are other solutions, yes, and my intention wasn't to say which way is better I was merely pointing out freely available software that can help reduce some problems people are experiencing.
 
Lee Wilson said:
Good point.

I suspect besides the (poor) choice of Motion Jepg - the way the image is extracted from the sensor plays a part.

Binning + old and inefficient codec on a low quality setting = poor image.
Click to expand...

Still, if we can get it before the compressor it will be nicer. Hardware hack I imagine. Otherwise there would be clues with this many eyeballs in frantic motion. ;)
 
@ Drew: Merlin stuff looks good! (Vimeo's playback is ag'in you—jerky.)

Exposure aside (it's fine), the skew looks acceptable to me. Thanks for this.
 
PenGun said:
Still, if we can get it before the compressor it will be nicer. Hardware hack I imagine. Otherwise there would be clues with this many eyeballs in frantic motion. ;)
Click to expand...

Yep ! that would be good, to extract the image before it gets mangled by MJPEG, but you would likely still be capturing an uncompressed signal with all its (suspected) binning problems.

By that I mean - circumnavigating the compression stage would solve only one part of the problem - although that would certainly be preferable.
 
Drew..that footage is great.
all you have done is follow the rules of decent filming (especially 24p), steady.

Thats enough from what ive seen in yours and zaks and kholis stuff with the
colour, Dynamic range, bokeh, (the skew/wobble is bad but not that bad from your last vimeo post)...ok not perfect then but stop thinking compression etc and look at the overall picture of what this camera can achieve.

Im seriously looking at using this now (and not upgrading my adapter).

There are some people involved in this thread who seem to do nothing but knock then slate then moan about this camera...and they just dont go away...do they own 35mm adapter manufacturing companies..LOL....eh?

Ok im probably gonna get crucified for that but..what the heck. ;)
 
Is this a good time to review what we’ve learned?

Is this a good time to review what we’ve learned?

I’m grateful to Kholi Hicks, Zak Forsman and Drew Ott for their fast D90 testing & uploading clips. I’ve followed this thread from its first day—read every heated word—& think I have a rough handle on the results of testing thus far. (The video clips in particular have been invaluable.)

So far the D90 looks like a qualified dream-come-true, for my purposes. But spending $1300 would require that I sell my XL2, so I plan to rent a D90 & put it through its paces to confirm it’s the right solution.

There are several technical points I’m unclear on. Rather than badger the thread with questions (you guys are all as busy as I am), I wonder if others would find it useful if we assembled & posted a SIMPLE review of what we’ve learned about the D90.

A review could be organized along the lines of “pros & cons” and “steps to achieve basic video functions.”

The functions might be:

1. Aperture, including a gloss on the kit lens vs. manual lenses
2. Setting ISO (i.e., gain), including a step-by-step of the “Kholi trick”
3. Shutter speeds, how to set them (or not)
4. utility (or lack thereof) of: HDMI out, saving to cards (what kind, etc) & AV-Out
X. [anything else that the experienced shooters here think useful]

Is it too soon to write up something like this? It might serve well as a guide for new thread visitors & guide further testing by guys like me with all the excitement, but fewer chops.
 
Last edited:
You must log in or register to reply here.
Back
Top