Mystery solved (sort of)!

12coyote wrote on 3/16/2012, 6:35 PM

To everyone that was following my "100% completely impressed" thread. http://www.sonycreativesoftware.com/forums/ShowMessage.asp?MessageID=805319&Replies=23

The below links take you to 2 identical videos, but they were rendered with different settings. To cut to the chase when a velocity envelope is used "smart resample" must be disabled. If you don't disable "smart resample" you will render a video much like the below linked "smart resample" video. Using a velocity envelope with "smart resample" disabled renders a video like the below linked "disable resample". Like yesterday's BMW M3R video today's Ferrari 458R Italia video is 1920 x 1080p @ 60 fps. Unfortunately, with today being experimental in trying to figure out what went wrong with yesterday's BMW M3R video the camera settings were slightly different yesterday to today. That being I slowed the shutter speed from yesterday's 1/8000 to today's 1/4000. Yesterday's OIS was set to HYBRID (optical and electrical) and today's OIS was only set to optical. However, in the final analysis I don't believe the camera settings will matter. I believe it's all about the "smart resample" and the velocity envelope as it pertains to 1920 x 1080p @ 60 fps video. Yesterday's thread was titled "100% completely impressed". With today's discovery I move to "110% completely impressed"!

Deleted videos from the server 3/17/12

Comments

farss wrote on 3/16/2012, 7:25 PM

Disabling resampling is causing frames to be duplicated. Kind of out of the frying pan into the fire. Resampling will make not do anything IF the frame rate that you're rendering to matches the source frame rate. That seems to be the fundamental problem that you need to wrangle.

"That being I slowed the shutter speed from yesterday's 1/8000 to today's 1/4000"

That's making almost zero difference. The skew is as bad as ever and as noticeable as ever. Changing shutter speed and the image stabilization does not change sensor readout time, changing shutter speed to get enough motion blur will mask the probem. Why are you using such a fast shutter speed?

TBH with such a fast shutter speed I suspect the camera is running out of light and having to increase gain apart from the other issues.

Bob.

12coyote wrote on 3/16/2012, 7:59 PM

It must be because you're working with the .mp4? I'm not getting any duplicated frames in the .mts. or the .mp4. I'm using a fast shutter speed in order to get a clear stop action. The lower the shutter speed the blurier it becomes when momentarily stopped. You're assuming that the frame rates are not the same in the project and the render. You'd be wrong in that assumption. I used the duplicate settings icon and it's 59.94 fps in both. I chekced it with each render.

farss wrote on 3/16/2012, 8:47 PM

"It must be because you're working with the .mp4?"

Well that's all you've provided that I downloaded from: www.davidjohnsonpage.com/disable resample.mp4

In V10 matching my project's properties to the source I have the frame rate as 59.940 (Double NTSC). Switching the timeline ruler to Absolute Frames I find frames:
80, 81
92, 93
96, 97
100,101
103,104
and more are indentical. Maybe your velocity envelope has some slope down to 0%.
That would explain how at the start there seems to be no duplicate frames and then the incidence of them increases.

"The lower the shutter speed the blurier it becomes when momentarily stopped"

That's very true although if your pan is exactly tracking the car then you'll capture very nicely the sensation of speed as the background and the wheels will be blurred but not the car. I appreciate that tracking a car on such a circuit to keep it still in the frame is no mean feat, certainly well beyond me even with an expensive tripod.
Still I'd think you could try slower shutter speeds and not get any problems with blur if you're pan is reasonably spot on.

Bob.

12coyote wrote on 3/16/2012, 9:05 PM

It's well beyond me as well. In reality well beyond any human being. Broadcast TV doesn't even try it on the ground level. They use elevated platform trajectory and cameras on gimbals.

farss wrote on 3/16/2012, 9:30 PM

It can be done, if you don't mind taking a lot of images to get something like this.

Agree, it's much easier from up high and for high profile sports broadcasts there's now cameras that run at something like 150fps to get good slomo playback.

Bob.

12coyote wrote on 3/16/2012, 9:59 PM

You're comparing a static image to a video? Huh? I have a ton that look just like that I took with a DSLR.

ushere wrote on 3/17/2012, 12:08 AM

30 years ago i was shooting close to trackside with a betasp rig on a satchler tripod.

tracking cars on tight curves wasn't much of a problem, other than nausea swinging backwards and forwards to catch the cars ;-(

maybe bob knows why there wasn't any problems with ghosting / skewing or the like in analogue (well, that i remember;-()?

farss wrote on 3/17/2012, 4:51 AM

"maybe bob knows why there wasn't any problems with ghosting / skewing or the like in analogue (well, that i remember;-()?"

That'd be because all the analogue cameras used CCDs as the early MOS imagers had multiple problems . The DV cameras also pretty much stuck with CCDs. It wasn't until the HD thing that MOS led to CMOS and the smear problem was solved that prosummer cameras started to go away from CCDs with their global shutter. The first HDV camera, the Z1, used CCDs so no roliing shutter artifacts.

The advantage of CMOS is less noise and better low light performance.

Bob.

amendegw wrote on 3/17/2012, 6:10 AM

Well, beauty is in the eye of the beholder. The "disable resample" footage is more pleasing to my eyes. This might not be the case at a lower fps. And, it might not be the case for other observers.

A while back, I started a thread, Whither Smart Resample where my last post is,

"I've done some more testing here and want clarify my recommendations - just so folks reading my comments are not lead astray.

So, what's the bottom line to the bottom line [grin]? Enjoy your new camera! You are taking some impressive footage. Ask questions to this forum, and go with what looks best to your eyes.

Just my opinion (and I guess, I'm stuck with it),
...Jerry

farss wrote on 3/17/2012, 6:44 AM

"So, what's the bottom line? There's no one answer for everything."

Absolutely. Even more confusing is you can get the right answer to the wrong question. Which as I've recently realised is worse than the wrong answer to the right question.

Bob.

amendegw wrote on 3/17/2012, 7:06 AM

Bob, Agreement! I love it!

"Even more confusing is you can get the right answer to the wrong question. Which as I've recently realised is worse than the wrong answer to the right question."

Forgive me for adding a touch of good natured levity, to an otherwise "dry" subject:

...Jerry

12coyote wrote on 3/17/2012, 9:45 AM

I love "Rummy" quotes! My favortie was his response when a stupid reporter asked him, "Why are we dropping those daisy cutter bombs"? Rummy replies, "Because we are trying to kill them". Only the late great John Mckay was better on the fly. When asked, "Coach McKay what do you think of your team's execution today"? Coach McKay replies, "I'm in favor of it"!

Andy_L wrote on 3/17/2012, 11:34 AM

So, what's the bottom line? There's no one answer for everything. Best advice is to do some test renders on your project and see what looks best.

What drives me crazy is that this also depends heavily on codec, plus bitrate, plus scene contrast, plus image stabilizer settings, plus frame rate, plus output device -- your specific tv or monitor, plus...who knows what else. And personal taste, too.

Maybe the hallmark of the digital age is that *nothing* is right anymore...

12coyote wrote on 3/17/2012, 11:48 AM

In my testing with 1920 x 1080p @60 smart resampling and velocity envelope do not mix under any circumstances. I think that is a constant in this digital age. At least until it's fixed in Vegas 12, or as an update to existing.