Horizontal Pans

[Stathis Athanasiou]

Hello,
I have been told that during a horizontal pan, there are compression problems and/or artifacts when in progressive mode.
I find that an exaggeration, but has anyone noticed something like that when horizontal panning, or during a high motion handheld shot?

Thanks

[Christopher Neville]

I use 30F at 1/60 for shooting waterfalls. I often do slow horizontal and vertical pans. I've never noticed any artifacts on my monitor when editing, or when reviewing raw or edited footage on tv.

[Chad Ream]

I have... when shooting in 24F. The image is great when static sitting on a tri-pod, but when I began panning it appears that their are little jaggies shaking, of course you only will see this in post.

I have chosen as of lately to film in 60i and have been very happy with the results for my purposes.

Chad

[Bill Pryor]

I've shot everything in 24f and have had no problems with artifacting.

[Raymond Toussaint]

Be sure though to turn of the OIS during panning...

[Ken Wozniak]

I think you may be referring to "strobing" of the image during fast pans in the 30F and 24F modes.

If you pan too quickly in either mode - more so in 24F - you run the risk of objects on screen moving so much between frames that they appear to strobe.

Keep pans slow and really lock on to the subject during the pans and you should have wonderful looking footage.

IMHO, the A1 produces beautiful footage in 24F and "hyper-realistic" footage in 60i.

[Kuentos San Miguel]

Can someone please help understand the me difference between 30f and 60i?
thanks

[Stathis Athanasiou]

Kuentos, I can't help you with that since I am in Europe and use PAL.
Ken, as for the strobing, I wasn't referring to that, but to compression artifacts. Like for example the "blocking" effect when you have a high motion sequence compressed to a low bitrate mpeg. I know that you would never see this as exaggerated as in the case of a low bitrate mpeg, but I've been told by a tech-freak friend of mine that only the HVX200 at 100MBps can handle well high motion shots or horizontal pans without artifacting.

[Chris Soucy]

OK, well, it's really quite simple.

There is only a finite data rate out of the camera to the tape.

If you exceed that data rate, something has to be dumped.

With HDV, with the long GOP system, if you do anything to provide more data to the system than it can compress and write to the tape, it gets dumped.

Take a flower. Have it in close up. Zoom back to include a vast expanse behind. All well, so far.

Now do a pan.

Every single pixel in every single frame has to change.

If you exceed the ability of the encoding to write that data to the tape, there is no way round it, it will dump the excess data.

In my experience, the one thing that will be maintained is the "central character" whatever that may be. All the rest will suffer quite noticeable degradation of the picture quality as excess data is simply dumped.

It is nothing to do with frame rate or anything else, simply a data rate block with HDV. It's probably the same with any other HD system that has to encode data, if you exceed it's capabilities, it has to do something with the excess - ergo, dump it.

Stathis,

Do Not do fast pans. Slow and easy will do it every time. If you must do a fast pan, make sure you have a fast moving target "front and centre" to take up the slack as the background turns to mush.

Most people do not see this degradation on their HDV footage as, suprisingly, they ain't watching it on an HD screen. When they do, it comes as quite a shock.

DO NOT EDIT HDV ON AN SD SCREEN!!!

It really isn't rocket science.


CS

[Mikko Lopponen]

Originally Posted by Chris Soucy

Do Not do fast pans.

Oh come on. I've been doing fast pans and fast motion for awhile now with hdv devices and they work fine. Even on an hd monitor.

[Christopher Neville]

Originally Posted by Chris Soucy

Most people do not see this degradation on their HDV footage as, suprisingly, they ain't watching it on an HD screen. When they do, it comes as quite a shock.

DO NOT EDIT HDV ON AN SD SCREEN!!!

It really isn't rocket science.
CS

Since I'm the only one in this thread that mentioned watching edited footage on tv, I get the feeling that the above statement is a passive aggressive swipe.

Just for clarification, I have a second computer monitor that I use with editing to preview footage, and my tv is HD plasma. I didn't believe I needed that level of detail for a quick response, but I stand corrected. If needed, I can provide model numbers also.

[Eric Weiss]

No, it’s not rocket science. The “rocket” has used the same science for decades while the HD landscape changes by the hour.

Panning is a combination of creative and technical skill. On the A1, 60i transitions fast pans with no problem at all. Too fast in 24 will strobe, but the image holds up. It’s nothing better or worse than what can be seen on TV right now.

If you want a really smooth wide pan in any frame rate, use a tripod and count to 5 within 180 degrees.

If you need to track a fast moving object, I would suggest 60i and use your skill to keep it in frame without excessive V or H camera shake.

I edit on a large computer monitor with both HD and SD screens simultaneously.

[Daniel Browning]

Originally Posted by Stathis Athanasiou

Hello,
I have been told that during a horizontal pan, there are compression problems and/or artifacts when in progressive mode.
I find that an exaggeration, but has anyone noticed something like that when horizontal panning, or during a high motion handheld shot?

What you were told is too ambiguous to judge for exaggeration, as it depends on how fast the pan is, how much detail is in the frame, and how closely you examine the results.

The last one is the kicker. It's easy to see the difference in 100% crops from frame grabs even on totally still footage (compared to uncompressed 4:4:4); but most are more concerned about the difference at normal viewing distances.

[Evan Donn]

Originally Posted by Chris Soucy

Take a flower. Have it in close up. Zoom back to include a vast expanse behind. All well, so far.

Now do a pan.

Every single pixel in every single frame has to change.

Actually, no - in that case every 16 pixel block simply takes a step to the right (or left, as the case may be). This is called a motion vector, and it's part of what makes MPEG2 such an efficient codec while still retaining high quality. In a pan, only the pixels along the edge of the screen where new detail is being revealed actually change and are encoded new; the rest of the image is simply encoded as a bunch of motion vectors saying move this block x number of pixels left or right. So a smooth, slow pan should not push the compression beyond the limitations of HDV's data rate.

Because of motion vectors it's important to distinguish between movement of blocks of pixels (panning on a relatively static shot) and actual changing pixels (people walking around, sparkling water, blowing leaves, etc) when determining what's likely to cause HDV to break up.

Now, if you pan too fast it's possible you could exceed the radius in which the encoder looks for movement of each macroblock, and that might cause some artifacts - but it's also likely that if you are shooting 24f panning at that rate will introduce judder that's far more distracting than the artifacts.

If your shot consists primarily of fine, moving detail (like a grove of trees blowing in the wind, or rippling water), and you are panning, the encoder can't use motion vectors because the pixels are changing too much from frame to frame, so you are likely to see artifacts.

If you are shooting handheld the movement is less consistent than a smooth pan on a tripod and you may limit the ability to make efficient use of motion vectors, but again this depends on how fast you are moving and how much detail is in your scene.

I'm not sure where the best trade-off is in terms of compression and shutter speed. Shooting 24f at 1/48 means that if you move the camera quickly you'll have a good deal of motion blur. Motion blur means you can't use motion vectors efficiently (because the blurry pixels don't look like the clear ones in the frames before) , but it also reduces detail right at the time when the encoder is working hardest and therefore may allow you to avoid noticeable artifacts during fast motion. A high shutter speed will eliminate blur, which means each frame has clear, highly detailed blocks that are easy to track motion vectors for - but it also means you potentially have a lot more detail per frame, and if those details are changing within the scene (i.e. not a largely static scene with camera movement) it may push the limits of the encoder. I suspect that all in all a slower shutter speed will be less likely to artifact, or at least produce less noticeable artifacts, but I haven't done any focused testing to prove this.

[Stathis Athanasiou]

Thanks everyone...
That was really enlightening!