XL-H1 = Great low light...No Grain.....Great for Videographers!

[Tom Roper]

Gold Star for Christopher. He's right.

Increase the gain setting for the Sony until you get the same noise as the Canon. Then compare.

[Shannon Rawls]

OK look, I think tis is the best way to solve this...

Since we are testing the "CAPABILITIES" of the camera to see in low-light.....

....then what we must do is simply create a dark shooting environment and I'm talkin' as dark as we can get it. And then see which camera can "SEE" the best in that place before 1-ounce of Grain is introduced in the image. If a smidget of grain is seen, back up to the next lower decibel until it's gone. Then compare the images.

How's that? Now THAT'S FAIR!

and also....record that same environment with all cameras on 0db as well. Won't take but a minute to do that too.

- ShannonRawls.com

[Michael Pappas]

I look forward to this day. Can't wait guys!

pappas

Originally Posted by Shannon Rawls

OK look, I think tis is the best way to solve this...

Since we are testing the "CAPABILITIES" of the camera to see in low-light.....

....then what we must do is simply create a dark shooting environment and I'm talkin' as dark as we can get it. And then see which camera can "SEE" the best in that place before 1-ounce of Grain is introduced in the image. If a smidget of grain is seen, back up to the next lower decibel until it's gone. Then compare the images.

How's that? Now THAT'S FAIR!

and also....record that same environment with all cameras on 0db as well. Won't take but a minute to do that too.

- ShannonRawls.com

[Johan Forssblad]

Originally Posted by Shannon Rawls

... it seemed like the Canon could see in the dark ...
- ShannonRawls.com

I did some none-scientific tests with the XL H1. Went out in the forest here in the darkest period during the year. The sun went down too early. Came finally to an old mill.

...

(I have nearly only been taking still photos before (with Hasselblad and Canon EOS-1n / D60). And one roll of 16 mm until I gave that up ...
I have waited with video until the quality should develop enough. Frankly, the resolution has been way too bad so far in my opinion.

I have driven twice through Africa from Sweden. Through the Sahara desert down in the rain forests. Sleeping on the soil for several months. Listening and photographing many of the animals. Wanted to project the great scenery and landscape of this wonderful continent back home. We were riding camels together with the touaregs. Carrying Hasselblad cameras because we considered the 35 mm SLR photos to be of marginal quality. We made slide shows with 4x4 m screen projecting 6x6 cm slides from Hasselblad projectors.

When comparing these slides with SD video projection the video looked as it was slides from a Kodac Instamatic! Not usuable in our opinion for great detailed landscapes.

Now with HDV the time has come for affordable video in our opinion. The moving "film" adds so much that we can sacrifice the rest of the resolution and sharpness.
...

Based on previous still experience I assumed I was out too late to capture any shots at all of the mill with the running water.

Everything looked grey and dark. Anyway, I focused on the water wheel (please forgive me - I am no native English speaking) under the mill house. Increased the gain to 6 or 9. Hey, many details came alive!

Tried some shots. Even from the wall in the shadow at the backside of the house. The sun was below the horizon half an hour or an hour ago but some snow helped with reflected light.

Looked at it on the computer back home. It was shot in SD only because a lack of computer programs for HD.

I was impressed! Somehow it nearly looked as it was taken in daylight. Many details on the wheel under the house was visible. I remember that I could barely see them by my naked eye.

Anyway, with my still cameras I think I should have headed home without taking any shot and returned another day to get a decent photo.

With this great camera some of these cuts could have been used in a video. And I didn't have the slow shutter on. So there is more potential.

Conlusion: This camera impresses me (but I do actually not have any other HDV or better professional camera to compare with).

I'm sure all experts here will put these findings into numbers later. Good luck!

[Mike Marriage]

Originally Posted by Shannon Rawls

OK look, I think tis is the best way to solve this...

Since we are testing the "CAPABILITIES" of the camera to see in low-light.....

....then what we must do is simply create a dark shooting environment and I'm talkin' as dark as we can get it. And then see which camera can "SEE" the best in that place before 1-ounce of Grain is introduced in the image. If a smidget of grain is seen, back up to the next lower decibel until it's gone. Then compare the images.

You could set the cameras up to record the dark scene, ALL SET AT 0dB. See which gives the "brightest" picture (i.e. which is most sensitive without gain) and then use gain on the other 3 to bring them to the same level.

Then when judging how they perform you factor in noise level. As we don't know what the DSP is actually doing, the gain setting is irrelevent - only noise level and sensitivity matter.

That would mimic a real life shooting situation best IMHO.

[Ashley Hosking]

Originally Posted by Shannon Rawls

....then what we must do is simply create a dark shooting environment and I'm talkin' as dark as we can get it. And then see which camera can "SEE" the best in that place before 1-ounce of Grain is introduced in the image. If a smidget of grain is seen, back up to the next lower decibel until it's gone. Then compare the images.

How's that? Now THAT'S FAIR!

and also....record that same environment with all cameras on 0db as well. Won't take but a minute to do that too.

SOUNDS FAIR TO ME!!! IMHO *smile*

[Alister Chapman]

The 0db setting on most cameras is often a compromise between noise and dynamic range. Manufacturers will have the CCD and processing set up in such a way as to get the best dynamic range from the CCD's with an acceptable amount of noise. Most professional camcorders have a -3db setting, which does decrease the amount of noise BUT you will often find that there is a slight trade off with the dynamic range (contrast handling) becoming slightly reduced. This may or may not be why the H1 has a -3db setting, but at least Canon have given the user the option to choose sensitivity or noise.

If you look at most cameras with anything other than 0db gain you will also normally see the contrast and highlight handling drop away.

[Brian Petersen]

I can't wait for the comparison test! Especially since there's all this debate going on with the HVX and the noise in low light footage we've seen. Any ideas on when we'd be able to download the test footage?

Will there by any tweaking with the "detail" settings on these cameras? Barry has said that taking the detail down to -5 on the HVX takes the edge off the noise making it less noticeable, but the noise is still there, obviously, so those settings shouldn't be changed on the tests?

[Bob Grant]

I'm kind of surprised that the noise issue doesn't get more attention. Noise is the single biggest issue with current content delivery technology. ANY noise is using up bandwidth, run out of bandwidth and the noise starts to look worse, much worse.
Using DVCProHD quite nicely avoids the issue during acquisition, your footage will not fall apart from the noise. That's all fine upto a point. The point where you have to deliver the content. Not too many viewers are going to be watching the content off a DVCProHD deck. More likely they're going to be watching a DVB broadcast or SD DVD or at some future date a HiDef DVD of some form, all of these use temporal compression schemes and as your footage goes through that process bingo, the noise starts to look worse.
I don't know about anyone else but I find jittery noise a real killer, it mostly shows up in the lowlights, not usually where you want the viewer watching but given how our eyes/brains work we're drawn to movement in the periphery of our vision and the result is we find it hard to remain focussed on the action, our eyes are drawn to the unintended stuff jumping around in the lowlight areas of the frame.
To date all low light footage that I've seen from all the HDV cameras has WAY too much noise to be acceptable and that's hardly surprising given the size of the CCDs, to do better would require a true miracle as you'd need to rewrite the laws of physics. Probably the HVX200 is going to be the worst of the bunch, that's to be expected, progressive scan CCDs result in more noise, it's all a compromise and a very tight one when you start off with 1/3" CCDs.

[Shannon Rawls]

Bob,

I guess Canon is smarter then people think.

I wondered why they would seemingly go 'backwards' in technology with an interlaced sensor for their new expensive flagship camera right after they successfully mastered a progressive sensor in the XL2. Didn't make sense AT FIRST. But now it's starting to become clear.

Them guys are over at Canon are some cold-blooded geniuses.

- ShannonRawls.com

[Shawn Alyasiri]

Shannon,

Can you report how they compare when you've zoomed in all of the way?

I've got the Z1 and was quite fond of it's low light capability (for HDV that is), certainly compared to my HD100. However, the Z1 loses a bit when zoomed in (which I thought was f/2.8). To use it on my second mini35 adapter, I need to zoom in almost all the way, which puts it around f/2.6.

As I read the specs on the Canon, I thought it was f/3.5 zoomed in, so even slower, but certainly more zoom capability than the Z1.

So, I'd be curious how it's doing in a lower light scenario when zoomed all the way in (presuming f/3.5, rather than f/1.6 wide), and what your confidence would be in taking it into a lower light venue every once in a while, without lights (live event, say a concert, wedding balcony, etc). Having the nice gain structure you reported would certainly help things...

Lastly, can you confirm the gain setting capabilities as it's currently shipping - is it every 3db (-3, 0, 3, 6, 9, 12, 15)? I thought there were gaps in the numbers on the spec sheet, and would like to know what you thought of the picture at higher gain levels, while still zoomed in. I hate gain like everyone, but sometimes its a necessary evil, and would like to know what the threshold levels are for decent pictures. I've used the Z1 around 12db in a pinch, and been very impressed - don't care for much gain on the HD100.

Thanks very much for your reports,

Shawn Alyasiri

[Barry Green]

Originally Posted by Brian Petersen

Will there by any tweaking with the "detail" settings on these cameras? Barry has said that taking the detail down to -5 on the HVX takes the edge off the noise making it less noticeable, but the noise is still there, obviously, so those settings shouldn't be changed on the tests?

Well, there are two ways to test -- you can test based on the defaults, but that doesn't really tell you much other than how the manufacturers set up their defaults.

Or you can test by matching each as best you can to a certain standard. Such as the detail/edge enhancement. Pick an optimal point that you're aiming for, and then attempt to calibrate each camera so that it's delivering comparable results to the other. That way you're measuring real resolution, rather than artificial sharpening.

Same with noise, like I said earlier: brightness with a comparable noise level is probably what we're wanting to see, so a 0dB gain test on each is valuable in and of itself, but the more important test, for low-light considerations, would also factor the effects of noise and gain in.

If I were to conduct such a test, I would set all the cameras to deliver comparable results, and have multiple sets of eyes verifying what they're seeing. I'd also like to have an advocate of each camera there to make sure that the best is being gotten from each unit. And for viewing the results, I'd have those same people watch the footage back, but not tell them which camera the footage came from, so they'd have to give their opinions completely brand-free, just based on the image they're seeing on the screen.

[Mathieu Ghekiere]

Originally Posted by Barry Green

And for viewing the results, I'd have those same people watch the footage back, but not tell them which camera the footage came from, so they'd have to give their opinions completely brand-free, just based on the image they're seeing on the screen.

Great idea, maybe if they can do the test, they first can get the footage up without saying which footage came from which camera, and the first days or week people can guess which footage came from which camera :-D
Then, the winner can get a price or something, the best camcorder :-D

[Ashley Hosking]

Originally Posted by Bob Grant

Not too many viewers are going to be watching the content off a DVCProHD deck. More likely they're going to be watching a DVB broadcast or SD DVD or at some future date a HiDef DVD of some form, all of these use temporal compression schemes and as your footage goes through that process bingo, the noise starts to look worse.
I don't know about anyone else but I find jittery noise a real killer, it mostly shows up in the lowlights, not usually where you want the viewer watching but given how our eyes/brains work we're drawn to movement in the periphery of our vision and the result is we find it hard to remain focussed on the action, our eyes are drawn to the unintended stuff jumping around in the lowlight areas of the frame.

It doesn't help that our Australian DVB HD channels are compressed mpeg2 crap. e.g. C7 dont even broadcast on "real" HD, they broadcast on 1280x576p. Since when did that become an HD format? The "quality" (noise) shows when I play the 7 HD Showreel on my 24 inch Dell 1920x1200 16x10 display...

Sorry for taking the thread off topic... Now back to running the sony cameras on 12db while the canon is on 0db and then comparing *smile*

[Shannon Rawls]

Originally Posted by Ashley Hosking

*smile*

(flipping through my papers) I don't see where you paid your licensing fee to use that Ms. Ashley!

- ShannonRawls.com
*smile*