There has been criticism before about the way Canon HDV cameras handle highlights and I have to say, based on my own tests, these comments are well founded.

From what I have seen so far, there is nothing eloquent about how the XHA1 handles highlights, even with the knee set to low. It goes from exposed to blown out in a single step.

What is a little confounding is that I always use zebra stripes to judge exposure and I've usually been able to record what I am seeing exactly how I want it.

I currently have the zebras set to 100, just like I had on my DVX and I am very careful about not letting certain areas clip. When I look at high contrast areas of my shots (for example, a tree on the horizon with a bright sky behind), I am seeing lots of clipping. This was clearly not happening when I was looking through the viewfinder, no zebras, good exposure. This should not be happening so I am thinking I will need to bump down the zebras to 90 or even 80 which seems weird to me.

Has anyone encountered this strange behavior?

If you shoot with zebras at 100 percent, the only time you should see the zebras would be on something white in the direct light part of the frame. Broadcast cameras I've used have always set zebras at 75-80% for normal use, and that's what I use on all cameras with no problem. With zebras at 80%, you see them on a medium sort of thing, like caucasian skin tones, light blue shirt, etc., or a gray card if you want that precision, then back off a bit, and you're usually there. If you're getting zebras at 100% on average reflectance parts of your frame, then you're overexposing.

OKay Bill, thanks for the info. Weird that I didn't have the same problem with the DVX. I'll make the adjustment based on your recommendation.

As a matter of interest how are you determining whether they are blown out or not? The reason I ask is that when using Vegas I've noticed that there's a lot of info above 100IRE (Using the Waveform Monitor) and if I pull that back down I get quite some detail back.

This may have nothing to do with it - I just noticed I had the issue with a shot with lots of sky in it and by using curves was able to get some detail back...but I'm not a professional shooter, yet, I know more about the post processing side. I've come from DSLR background where you 'shoot to the right' and pull the info back in using the raw data afterwards.

...maybe I'm on the wrong track here.

Simon, I noticed it because there were chunks of branches missing. Their ends were detached from the main tree because the intensity of the sky had eaten them up. There is no rescuing this information because it was never recorded to tape.

It's rare when you can get any information from an overexposed area like this.

I've not noticed this at all with my XH-A1.

Can you post a screen grab please and I'll try and replicate.

Cheers

TT

There are some electricity pylons on the horizon against a bright sky. My zebras are on 100 and even when the sky is uniformly registering at 100% the pylons are still sharp. It is not until I over-expose by about 2 stops that the detail is lost as you'd expect.

In your shot are you shooting into the sun causing the tree to be in silhouette?

I've always thought the XH-A1 handled highlights pretty well. As good as any other current compact HD camera.

Tony, I think the problem has been identified. I just had the zebras set too high. A simple adjustment should fix the problem. I'll let you know.

Thanks

Working with zebras set 100% is risky by definition. Even without reaching the limit, brightness levels at some spots in conjunction with all factors at play (starting with haze and air quality, through lens properties, ending with electronic path and codec limitations) may cause problems that you have noticed.

To really figure the difference between A1 and DVX behavior you would need to compare then side by side in the same conditions. I don't think there's much of difference here.

All cameras, film and digital, suffer this way because dynamic range of typical daytime scenery exceeds their capacity. It's only a question of pushing them far enough. The other reason is limited resolution of zebra and sometimes one may not be aware that clipping actually occurred. That's why most people prefer to work with zebras set less than 90%.

Hi Steven,

I noticed that too, early on when I got the camera and set the zebra's
to 80%, that helped a lot.

good luck with it,

David

Okay, thanks everyone...I'll do some further testing using 80% zebras but I'm pretty sure that will work.

Oops, some server or connection problem occurred and my previous post got doubled.

Chris - could you do a favor to remove this one? Thank you!

I don't understand why 100 zebras would cause the problem. 100=100 as far as I can see.

I've not had a problem with Canon Zebras set at 100.

I don't understand....

TT

Okay so here's the deal.

I realized when I looked at my settings on the camera that my zebras somehow got turned off so I was assuming they were on and looking to see them when I was shooting this morning. I assumed I was within exposure range but was not.

So that's my embarrassing confession :)

Nonetheless, the Canon cameras do blow out very quickly even with a low knee setting and it will suit me better to have an "early warning system" by putting the zebras on probably 85% or 90%...

The stinker with the A1 is that if you happen to toggle peaking on and forget about it, you've lost your zebras. It would be really nice if you could use peaking and focus assist while the cam was recording...and still keep the zebras up!

Steven, I've noticed it too, the A1 is quick to blow out, particularly as compared to the dvx. I keep peaking on due to the weak LCD, and I miss the zebras. If you're using the lcd, have you knocked down the brightness? It helps somewhat.

I like to shoot a lot into the sun/light source for a blown-out, overexposed look for some pictures. Under these conditions it's pretty much impossible to avoid the zebras going mad, even at 80%, and indeed parts of the picture will be overexposed. But I'm from a still background...

Does it work to shoot like this, then pull the levels back in post, so that it's broadcast legal, even though it will still look blown-out, because that's the look I'm wanting?

My understanding is that most consumer and prosumer video cameras record into the 109% superwhite range.

So theoretically, with zebras set to 100% you should still have a little lattitude and a little retrievable overexposure. (I seem to recall on my DVX there was a 105% zebra option, so you could just flag what was gone for good!)

In reality I think it probably depends on how the processor works. I've only been working in controlled lighting situations with my A1, so I haven't had a chance to experiment with this one. Perhaps 100% zebra with the Canon is actually 109% IRE [sic]. Or perhaps it records up to 109% but blows out after 100...

Stephen

What gamma setting are you using? I can definitely see an advantage to setting the zebras lower if you are using Cine2 gamma curve as it starts to blow out highlights well before they actually get to 100 IRE.

My setup for those that are interested.

I have found the NORMAL gamma curve give me the best highlight handling and the look that CINE1&2 give can easily be replicated in post without destroying the dynamic range while shooting.

I have the BLACK STRETCH and KNEE LOW. SETUP also affects the gamma curve and thus the highlights.

I also use a modified NOMAGENT preset downloaded from the presets section.

I find this gives me a wide dynamic range and neutral look ready for achieving the desired grade in post.

I am not suggesting this is the way to work just the way I like to work with the camera.

TT

Tony,

A very good point about the gamma setting influencing actual latitude! I can confirm this on both the A1 and Sony V1. The zebra display doesn't account for that, and thus can be misleading (especially when set at 100%).

So far I haven't worked with any prosumer camera that really does a particularly good job of handling highlights.

And so far the improvement with CMOS based cameras is only marginal at best.

Normally I feel like I really want to be able to set the knee a bit lower. On the A1 I can't really see any difference between the different knee settings.

And so far the improvement with CMOS based cameras is only marginal at best.

Normally I feel like I really want to be able to set the knee a bit lower. On the A1 I can't really see any difference between the different knee settings.

Indeed, the V1E does not offer any more latitude than the XH-A1 and handles highlights no better. The real benefit of CMOS is to be able to adress each pixel and make changes on a per pixel basis. The EIP in the V1 adds a bit of faux latitude with the contrast enhancer option as it lifts the blacks but at a cost of more noise.

Back to the XH-A1, I can clearly see a difference between knee settings. Good places to spot the differences are on bright curved surfaces.

TT

Back to the XH-A1, I can clearly see a difference between knee settings. Good places to spot the differences are on bright curved surfaces.TT

Oh yes. Now I see it. Not quite as big a difference as I'd like... but every little helps.

So, NORMAL gamma seems to give a wider dynamic range than either of the CINE gammas...?

What about SETUP to -9 and PEDESTAL to +9. It seems to my eyes to give a slightly flatter wider range, without washing out or crushing the blacks.

It is possible different manufacturers define maximum level differently. Maybe zebra 100% is what Canon defines as the peak level, maybe Pana defines it somewhat different. The key is we need some time to experiment with our cameras to learn their characteristics.

CMOS does not offer any significant advantage over CCD's dynamic range. It's the same photodiodes that receive the light, only addressing, signal transfer and integration is different.

Phenomenon Steven noticed is quite common and often unavoidable. I'm sure we all have seen it working with film or digital cameras. Even RED will suffer the same way when light is tough enough.

There's a new high dynamic range technology slowly emerging that uses sensors with special matrixes delivering 2 levels of response to light. As a result, their dynamic range is similar to human eye. It does not look like this technology will be commercially available in this decade. Maybe in 5, maybe more years from now.

Undoubtedly a (slightly) better job can be done now, as there are high end cameras with more sophisticated DSPs that handle highlights more elegantly.

But I guess it's partly cost of technology; and partly market segmentation.

And I look forward to any technology that can improve the dynamic range.

Undoubtedly a (slightly) better job can be done now, as there are high end cameras with more sophisticated DSPs that handle highlights more elegantly.

But I guess it's partly cost of technology; and partly market segmentation.

And I look forward to any technology that can improve the dynamic range.

Of course, even SD cameras handle highlights slightly better because their photocells are larger.

HDR? Oh man, simulating it with still images I just can't wait to see that dynamic range in moving pictures! Until then, I dial Knee Low and use bounce board and screens whenever possible to balance the light better :)

I seem to remember Adam Wilt's write up on the shootout and the XL-H1 was the camera that handled highlights most elegantly. Canon ensures there is no hue shift towards the highlight.

TT