FX1 and Sports

[Mike Hanlon]

How does the FX1 (and the HDV format in general) fare with fast moving sports images? MPEG-2, in my experience, doesn't handle the fast moving imagery of sports very well. My football highlights video shot on DV looks great on tape, but disappoints when played from DVD.

Am I going to be similarly disappointed by HDV?


Mike.

[Kaku Ito]

Yes. Especially with back ground with a lot of details. I shoot downhill mountainbiking and i can't pan a rider passing in front of me without the resolution loss.

Take a look at this.

http://homepage.mac.com/kakuito/blog...28/index.html#

[Ben Buie]

Please note that the referenced mountain biking shot doesn't really have anything to do with the fast movement of the biker; it is simply that because Kaku was close to the action he had to whip the camera around to follow the action of the rider.

In other words, it is not fast movement that causes a problem, but rather fast movement of the camera itself.

The problem occurs because the camera needs a frame of reference to base the I-frames on. When the CAMERA is moving fast, it loses that frame of reference. The biker (or in your case football players) can be moving like crazy, and it wouldn't be a problem, as long as either the camera is steady or panning steadily. In other words, a tripod and/or hand-held steady tracker is highly recommended.

The best idea is to set up your camera as far from the action as possible, and use the 12x zoom as needed. You will also get shallower DOF that way.

[Kaku Ito]

Ben, thank you for your notes, but I'm afraid that my conclusion is based on the results of after trying everything you mentioned.

I even wish field video shooting is as sorted out as you described. My overall impression that I got from shooting many sports oriented events after October 2004 with FX1, having such limitation really hinder you in the shooting situation at a spot.

If you place your camera far away from the mountainbike course, along the course will be filled with gallery, you won't even catch the rider completely. I'm into riding, too, so that I know what it takes to go down fast in downhill mountainbiking and what I want to express in my video is not really far away shot without the closeness of the video and the sound.

You can get around with shooting nice picutre, but I really think technology should not limit what you want to express. I'm trying my best to cover this weakness of HDR-FX1 with carrying a sub-camera (NV-GS400) and shoot close shot with it (but it takes another person or it can only shoot boring fixed point view) or using my glidecam or using very light but feasible video tripod (Manfrotto 503 head and carbon leg tripod), and I even have Glidecam V8, I am really working hard to bring the best of FX1, but my conclusion at this point is that if you really want to express what is happening with the sports events, FX1 is far from ideal. I have to caryy so much more extra gear, think and plan a lot for getting satisfactory footage. Ofcourse the scenaries and people images come out really nice, but unexpected sudden movements of athletes' demand a lot more consistant performance.

I will be further challenging the use of FX1 my field of videography. I'm currently writing a column in a Japanese magazine about utilizing FX1 in such videography, so I will try to provide my findings as I progress along.

Thanks again, and readers can decide what to do reading your (Ben's) work around and at the same time, think of how much hustle it is to be limited in way of shooting.

[Barry Green]

The question comes up because of the nature of MPEG-2 compression. The amount of "action" isn't necessarily what determines whether the MPEG-2 will be able to "keep up" with the image. It's the amount of changes per frame.

The more that changes from frame-to-frame, the more MPEG-2 is challenged. And it's possible for there to be too much change, at which point MPEG-2 cannot compress all the detail, and the result is loss of resolution.

Someone else on another thread spelled it out pretty well -- it's not what we think of as a lot of action that matters, it's what the camera sees as a lot of change. So Ben's suggestion would work (as he can testify from actual experience) that a locked-down shot with a mountain bike going through it would be within MPEG-2's ability to resolve with adequate detail, but Kaku's experience is that if you go in and hand-hold, the MPEG-2 cannot resolve it adequately -- because in Kaku's situation, the BACKGROUND is also moving, which means the entire picture is changing, which overloads the codec.

Another example that was given was a static shot, but with "heat waves" causing ripples in the picture. In that shot we, the viewer, would perceive no "action", yet the heat ripples would cause nearly every pixel on the screen to change, and that could overload the codec.

Another example I saw was with the HD1, a guy shot a golf course sprinkler (one of those big water-blasting sprinklers). In 30P on 1/30th shutter, the picture looked gorgeous, filmlike and beautiful, sharp detail, etc. But when he switched to 1/60th the picture looked horrible -- it totally fell apart. The difference is that in 1/30th the water droplets were smearing and streaking, whereas in 1/60th they were more crisply resolved, and the amount of crisp detail overloaded the codec.

DV (and its variants, DVCAM, DVCPRO, DVCPRO50, and DVCPRO-HD) do not suffer the same type of degradation in these circumstances, because all compression is contained within each frame, and every frame gets the same amount of "bits" allocated to it. But MPEG-2 doesn't keep all compression within one frame, it groups frames together and makes each frame rely on the first frame in the group. It's therefore possible for MPEG-2 to "run out" of available bits, as it were, and then have not enough bandwidth to deal with all the high-detail, high-motion frames... and without enough "bits" to go around, the result is lower-resolution frames (similar to how web compression used to look).

It would be nice if we didn't have to worry about things like this, but unfortunately, if you want to shoot HDV, you have to be aware of it and perhaps adjust your shooting style to accomodate. When 24P first came available on prosumer-priced cameras, people found out pretty quickly that they either had to adjust their shooting style, or they'd shoot stuff that was not visually appealing. The 24P frame rate can't be shot with the same freedom as 60i could. I believe that as people get more experience with HDV we'll be able to share techniques that deliver the best results, and techniques that usually lead to issues, and we'll all be better able to recognize the situations that might cause problems and find ways to avoid them. But there will be a learning curve.

[David Kennett]

Mike,

Making a DVD of your interlaced DV footage should not affect the smoothness. Your MPEG encoder should give you a choice of field order (A,B or B,A). Having this wrong will make for very jerky motion. Since the FX1 is interlaced, this will still be a problem. On the other hand, the HD 10 progressive makes things not quite so smooth, but there is no chance of getting the field order wrong when converting to SD.

[David Kennett]

Mike.

One more thing. The difference will NOT be apparent on your computer's progressive display - only on regular TV with regular DVD player.

[Mike Hanlon]

Thanks for the thoughtful answers.

I believe my answer is basically "Yes", the FX1 will suffer from the same kinds of problems that my DVDs do, specifically in my case I can expect pixelation around objects that change position from frame to frame. MPEG-2 compression does not handle fast changing scenes very well as it counts on being able to record the _deltas_ from frame to frame and not every frame, to save space. As it was put by Barry, MPEG-2 "runs out of bits" when too much changes from frame ot frame.

That said, I can't say that I see the same kinds of pixelation problems when watching (action) feature films on DVD. Is that because film is "softer" and therefore doesn't put such a demand on the compression or is it something about the encoders they use? Is there a way to configure my encoder, (Compressor) to use more bits and compress less? I'm already using 2 pass VBR encoding with a maximum bit rate of 7Mb/s.

One other peeve of mine with regard to MPEG-2: why can't it handle a simple .5 second fade up from black without banding and pixelating? Again, is this operator error on my part?


Mike.

[John Jay]

<<<-- Originally posted by Kaku Ito : . I'm trying my best to cover this weakness of HDR-FX1 with carrying a sub-camera (NV-GS400) and shoot close shot with it . -->>>

Kaku - try this

record with FX1 as normal and tape the S-video feed from the FX1 as DV on the GS400

The S-video feed should be free of m2t encoding since it is from the camera head and should record on the GS400 very well so long as you use a quality cable

[Kaku Ito]

John Jay,

Never realized that! I guess on the GS400's recorder mode should record incoming video signal as anamorphic, so it should be okay.

I will try this and report it on the magazine. I will credit your name.

One thing we have tried is that we connected the component out of FX1 to analog in of the new Decklink Multibridge, convert the signal to HDSDI and record with FCPHD with Blackmagic Decklink HD Pro at uncompressed 8-bit HD/1080i. It is beautiful but exprerienced consistant problem that when the camera was panning very slowly from something shinny (chrome color metal) to subtle object (it was my white leather turtle), the signal goes off, so I would get intermittent video. Anyway, anyone interested in such video footage?

[John Jay]

Kaku

Hope it works :)

Yes please would love to see some component :)


www.yousendit.com can handle 1 gig files!

[David Farland]

You probably already know that tech support guys from Blackmagic Design co-host the Decklink forum below. I'm sure they'd be more that willing to help you with signal dropout.

http://www.creativecow.net/forum/vie...hp?forumid=124

Dave Farland,

ps: I'm investigating solutions for capturing the FX1/Z1 'live' component output on a portable PC system.
Needs a few technology changes first to make it easily achieveable.

[Mike Tiffee]

I'm working the host show for ABC at the Sugar bowl today. We're done at halftime so I'm going to take the FX-1 inside and shoot some footage. I'm going to shoot some footage at high and low shutter speed, CF 30, CF 24, no CF. If somebody has some server space I'll upload some 10 second clips or something.

[John Jay]

<<<-- Originally posted by Kaku Ito :

One thing we have tried is that we connected the component out of FX1 to analog in of the new Decklink Multibridge, convert the signal to HDSDI and record with FCPHD with Blackmagic Decklink HD Pro at uncompressed 8-bit HD/1080i. It is beautiful but exprerienced consistant problem that when the camera was panning very slowly from something shinny (chrome color metal) to subtle object (it was my white leather turtle), the signal goes off, so I would get intermittent video. Anyway, anyone interested in such video footage? -->>>

Hey Kaku,

any chance of upping some uncompressed component?

A few secs will do nicely so I can apply some ADI to it. :)

www.yousendit.com can handle 1 gig files!

[Kaku Ito]

Okay