HV10/20 and rolling shutter

[James Bresnahan]

I'm curious if any HV-10 owners have experienced "rolling shutter" artifacts.

This was a big issue with the HC-1 when it first came out. Basically, since the rolling shutter system was unable to expose the whole CMOS sensor instantly, you sometime wound up with slanting and distortion on fast moving objects
(sorry if I've over-simplified the problem).

Not to say anyone tossed their HC-1 because of this, but it was in fact an issue for me, most notably in event videography when flashes occured from other people's cameras. The Hc-1 never recorded a full "flash frame"--always random strips of bright and dark in the frame, even at 1/30 and 1/60 shutter
speeds.

I'm just hoping, with an HV20 on pre-order, that this camera's shutter (or the Digic DVII chip) avoids this problem.

From the lack of complaints about this issue in this forum, maybe it is much ado about nothing on my part.


Jim

[James Bresnahan]

No worries, then?

I'll follow up when I can actually test my HV20.



Jim

[Robert Ducon]

I'm still getting an HV20 ;) But yes I wouldn't mind knowing as well. I agree, the lack of comments about this with the HV10 strikes me as a "good" thing.

[Fergus Anderson]

I believe the HV10 does suffer from rolling shutter. I have an example I can post if you like in which I did a fast (horrible!) pan past a lamp post which on the footage looks bent! I think the effect is made worse when using slower shutter speeds (mine was at 50 I think) and when the object is nearish to the camcorder. Im not sure if the progressive recording option on the HV20 will remedy this?

[Barry Green]

The HV20 manual addresses the "rolling shutter" question in the "troubleshooting" section. Basically they say that yes, on a horizontal pan, vertical objects may look like they "lean".

[Glenn Thomas]

I think this is the only real issue holding back the HV20 from becoming truly great camera. From what I've seen of the footage posted up so far, it looks great. Still, I'm looking at selling my HC1 to buy one of these to use until I can afford an XHA1.

[Alex Leith]

As I understand it the rolling shutter effects all CMOS based cameras. I think it may be one of those things you have to "work" with.

[Joe Busch]

I noticed it in some of my footage, it's only really really noticable if the camera is moving fast and the shutter speed is high (running at 1/500) I could notice it a lot more than at 1/60

[Rick Llewellyn]

An interesting, and unpleasant aritifact of the rolling shutter can be caused by a photo flash. If you are shooting with the HV10 at a time when other people are taking still pictues with a flash you will get the flash split across 2 frames. (not sure how fields enters into this). I was recently reviewing some video when a flash had gone off. The top part of frame 1 was clear and correct, the bottom part was washed out from the flash. On frame 2, the top was white and the bottom OK.

The effect of the segmented image seem a bit more jarring that what you expect from a CCD I found it quite annoying.

I saw this effect while viewing the video in FCP, so I am not sure if the visual impact would be less viewed on an interlacted TV.

Rick

[James Bresnahan]

As promised, here are frames from an HV20 showing what happens
when recording a strobe light (my rebel XT flash) with a rolling shutter.

For comparison, I recorded the same frames with an Elura, which is a
CCD camera and uses a different shutter system. You can see the Elura
records a clean full frame of strobe exposure. The HV20 cannot do this.

It seems this is an annoying limitation of CMOS camcorders for now. That said, however, I am so thrilled with capabilities of this camera, and see Canon bringing more of its larger frame Cmos technology to consumer videography.

[Povl H. Pedersen]

One simple question. At what shutter speed was the video with the black bars taken ? It looks like it is probably the first shutter faster than the speed at which there is no problem (if you are using a flash).

Please try again with faster shutter (larger black bar) and slower shutter (no bar I hope).

[James Bresnahan]

Originally Posted by Povl H. Pedersen

One simple question. At what shutter speed was the video with the black bars taken ? It looks like it is probably the first shutter faster than the speed at which there is no problem (if you are using a flash).

Please try again with faster shutter (larger black bar) and slower shutter (no bar I hope).

Yes, I realize this needs to be tested w/ a range of shutter speeds.
I will try to post more images covering this.

For clarification, in the currently posted images, the Elura's shutter speed was set to 1/60.
The HV20 was 1/48 in 24P mode. The flash was in pre-flash mode (red-eye reduction) where it flicker-strobes for about eight video frames.
All of the Elura frames were cleanly illuminated by the strobe, while most of the HV20 frames had the black bar at various frame heights.

As I said, I will try to post more accurate data, because this might be of
interest to anyone shooting events where flashes or strobes occur, but I also don't want to get too overtly focused on a technical limitation that is invisible 97.5% of the time. I'm still quite enamored of my HV20. ;)

If I may throw this in, you know what I like most about this camera (after flawless 24p)--its the reds. After 1 1/2 years with a sony HC-1, I love that red is red.




Jim

[Dave Blackhurst]

UGH... I've seen "rolling shutter" on an HC1 (flash on top and bottom of consecutive frames, only really noticeable if you're scrolling frame by frame), and the stretch/tilt if you move your cam around too fast...

BUT those black bars are horrid! What settings was the HV20 at to get that? It looks more like a bad head clog than "rs" - I'm sure it isn't, but it would make any footage unusable - not a good wedding camera if this is how the Canon handles flash... can't imagine a glitch like this got past engineering.

How many frames are affected? Just one, or several? IOW, in normal playback does this show, or just frame by frame?

[Mark Patrick Anderson]

I get my HV20 today and I'm going to try and do some tests with a fan and discover at what point and what settings problems happen.

[James Bresnahan]

Originally Posted by James Bresnahan

Yes, I realize this needs to be tested w/ a range of shutter speeds.
I will try to post more images covering this.

For clarification, in the currently posted images, the Elura's shutter speed was set to 1/60.
The HV20 was 1/48 in 24P mode. The flash was in pre-flash mode (red-eye reduction) where it flicker-strobes for about eight video frames.
All of the Elura frames were cleanly illuminated by the strobe, while most of the HV20 frames had the black bar at various frame heights.

As I said, I will try to post more accurate data, because this might be of
interest to anyone shooting events where flashes or strobes occur, but I also don't want to get too overtly focused on a technical limitation that is invisible 97.5% of the time. I'm still quite enamored of my HV20. ;)

If I may throw this in, you know what I like most about this camera (after flawless 24p)--its the reds. After 1 1/2 years with a sony HC-1, I love that red is red.




Jim

Here is a link to a mt2 file showing the DSLR flash strobe recorded at various
shutter speeds at both 24pf and 60i. The clip starts off with an older
CCD miniDV camera (Canon Elura) recording the same event. The Elura records the event cleanly. The HV20 records hard banding at every shutter speed, even the slow speeds. This still is not a scientific test, but I think it
demonstrates that clean records of standard camera flashes are not likely
with this camera--another limitation of the Cmos rolling shutter. I did get a
clean flash once in 60i at one-thirtieth shutter, but I could not repeat this--it
seems to be due random timing luck.

I guess some math could be done to correllate the "rolling" shutter timing with
the adjustable electronic shutter, to fully explain the banding patterns and
perhaps find a sweet spot for the cleanest recording of strobes and camera flashes.


http://www.sharebigfile.com/file/138...rTest-m2t.html

The file is 117mb mt2 reduced to 720 30p.



Jim