motion artefacts in fx1/z1

[Filip Kovcin]

i just saw something which looks strange/interesting. i grab the frame from test shooting from (in this specific case) fx1.

and saw that motion blur is not like in film. i.e. if someone is running , say from left to right - the blur ("ghost" tail) is BEHIND that person/ object.

but here - is BEFORE! it looks like that object is going BACKWARDS!

need someone's help to put that framegrab to server - i cannot right now access my own.

thanks.

filip

[Martin Doppelbauer]

HDV is Upper Field First while DV is Lower Field First.
Please check your NLE setup.

[Barry Green]

Was the footage shot in CineFrame 24? CF24 could probably create an artifact like that.

[Filip Kovcin]

<<<-- Originally posted by Barry Green : Was the footage shot in CineFrame 24? CF24 could probably create an artifact like that. -->>>

CF25.

but why?

is this connected with GOP?

[Barry Green]

CF25 shouldn't cause a problem.

CF24 generates frames using fields from different points in time (as described on Adam Wilt's site) and could theoretically create a forward-ghosting type of image.

CF25 shouldn't cause that.

[Filip Kovcin]

<<<-- Originally posted by Barry Green : CF25 shouldn't cause a problem. -->>>

barry,

but i can CLEARY see that there is ghosting image in front of moving object.

i will kindly send this to someone's server to analize - because it looks very strange.

filip

[Steve Crisdale]

<<<-- Originally posted by Filip Kovcin : <<<-- Originally posted by Barry Green : CF25 shouldn't cause a problem. -->>>

barry,

but i can CLEARY see that there is ghosting image in front of moving object.

i will kindly send this to someone's server to analize - because it looks very strange.

filip -->>>

I've noticed the same thing myself.... shot of a ferry crossing Sydney harbour. I was panning with it. The HDV component out playback on my 83cm HDTV looked fine, but when I downconverted to DVD (going from 50i to 25p) the strange forward ghosting effect appeared!!

My initial reaction was that it could well have been the predictive nature of the MPEG2 encoding used by the cam, but not every pan or higher movement sequences exhibited the effect..... only those with high contrast lighting. It's something I've been meaning to keep an eye on to confirm. Just hadn't got around to shooting in the same sort of conditions yet.

[Barry Green]

Yeah, that's odd...

Filip, I wasn't saying that you're not seeing what you're seeing, I was just saying that it shouldn't be due to the way CF25 works (which is a straight de-interlace, by all accounts) whereas I could definitely imagine that it could happen in CF24 because of the way CF24 works.

I'd love to see a still of what you're experiencing. Or do you only see it on motion footage? Can the effect be seen on a still?

[Filip Kovcin]

<<<-- Originally posted by Barry Green : I'd love to see a still of what you're experiencing. Or do you only see it on motion footage? Can the effect be seen on a still? -->>>

i can send you a still frame from motion clip of course, to your mail if you wish - i do not have possibility to make a link to this frame. i do not have access to my old site.

but - with your permition - will send it to you, it's full frame - tiff about 4Mb.

is it ok to send it or should i convert it to something lighter?

filip

[Barry Green]

Ship it on over! Send it to "barryg -- at -- icexpo -- . -- com"

Thanks!

[Dennis Adams]

What were you expecting exactly? Sharp in front and blurry in back, like a cartoon with "speed lines"? That's not how it works.

Motion blur is equal in both directions. Imagine taking a spray can and while moving, hit the spray nozzle for a second, and then release (all while still moving). The leading edge ramps up just like the trailing edge ramps down.

Same thing with exposing film or a CCD.

///d@
Sony Media Software

[Steve Crisdale]

<<<-- Originally posted by Dennis Adams : What were you expecting exactly? Sharp in front and blurry in back, like a cartoon with "speed lines"? That's not how it works.

Motion blur is equal in both directions. Imagine taking a spray can and while moving, hit the spray nozzle for a second, and then release (all while still moving). The leading edge ramps up just like the trailing edge ramps down.

Same thing with exposing film or a CCD.

///d@
Sony Media Software -->>>

You have an FX-1 Dennis? If you have then you'll get to see the effect at some point....

Try hand held shooting in a dimly lit interior....like a church....and shooting a window or on tripod in the same interior, shooting moving objects passing in close proximity or past the light source.

Then do a DVD downconvert (progressive) and you should see the bizarre behaviour....almost like a dancing shimmering of certain things while others remain unaffected.

[Barry Green]

Okay, Dennis probably cleared it up.

You guys are talking about something like when you shoot a panning shot of a white object (like a picket fence) against a black background, you'll see a thin white object, with translucent "ghost" white image in front of and behind it, right?

If so, that's completely normal behavior. The image is being exposed (and blurred) across the frame. The leading and trailing edges will be fainter, because the white image is being exposed across the full duration, but so is the black image (when you start panning, a certain pixel may start exposing all black, and by the end of the pan, it's started exposing white). So those pixels that only exposed white for a portion of the time will be "translucent", or "ghosted" -- and they'll happen in front and behind, because as far as the camera was concerned, the white image existed in both those places at some point. Now, the pure white pixels will come from the fact that the panned image saw only the central part of the white object -- the pan wasn't far enough to pan the white image completely off. So it saw white from the start of exposure to the end of exposure, and therefore registered as full solid white.

So what you're seeing, if this description makes any sense at all, is perfectly normal. Of course, the spray can analogy makes the point a lot better and in a lot fewer words!

[Filip Kovcin]

<<<-- Originally posted by Barry Green : Ship it on over! Send it to "barryg -- at -- icexpo -- . -- com"

Thanks! -->>>

barry, i just sent my frame grab to you,
please make link to it - if you can of course.

thank you.

filip

[Filip Kovcin]

<<<-- Originally posted by Dennis Adams : The leading edge ramps up just like the trailing edge ramps down.

Same thing with exposing film or a CCD.

///d@
Sony Media Software -->>>


dennis,

maybe i'm wrong, but i do not agree with you. maybe this is the case with spray, but (at least in my opinion) not with the moving objects in front of the camera/ccd.


spray is anamorfic in its nature. not solid!

when barry make link to my photo - you will see, that one part of object is less transparent - the other more. and the ghosting is somehow "predicted" - the object of ghosting is not yet on that position in the frame.

picture is shot without pan movement (but handheld camera).

and the WHOLE IMPRESION - when you look at the movement of all frames is that is some strange blur IN FRONT of the object.

i beleive that this is not my first ever movie i saw, but i newer saw something similar.
i.e. panasonic dvx100 is recording in somehow different way, and i cannot see similar ghosting on material shot with dvx100.


if you want - i will make another test with both cameras - sony FX1/Z1 and panasonic - and shoot same object. so then will be easeir to judge. (i hope)

filip