Who's used an HVX and has an opinion on (video) *noise*?

Elton said:
This is surprising to me as I've posted a lot of H1 footage, and noise was never a noticeable issue.
It's got good sensitivity, and it actually has active noise reduction capabilities, but frankly, for a green screen shoot, I'm trying to understand how this camera wouldn't have warranted consideration. It's SDI output is ideal for controlled studio shoots. I've piped SDI straight into my Mac with no issues, and recorded very good DVCPROHD 1080p with it. In all practicality, how would this be any different than pulling DVCPRO HD off of P2 cards?

Have you seen 24F material? The motion and resolution might surprise you...and you can acquire it in an array of quality codecs.
Click to expand...

That was an option we'd thought about, but for this shoot we were looking for something simple and self-contained. We didn't have a portable SDI deck nor an SDI in on our Xpress Pro HD setup.

There are actually many options, alterations to a DVX to get pretty good uncompressed video directly off the CCDs, recording the HD component out of even a pretty cheap HDV camera, etc. But we didn't have reason to believe there would be a problem with the HVX, and it was something we overcame.

Somebody else asked why we didn't use HDCam or something else given the nature of the production... it comes down to this being a very rushed and fairly low-budget couple shots for a very limited distribution. As we normally work in scanned film we don't have any HD decks in-house nor is our little Avid particularly well setup to capture from one. Because of the P2 workflow we were able to start working on footage as the shoot was still going on. So renting the HVX worked out OK. Truth is, in the past when we get up to that kind of price point and have the time we've generally rented a film camera. But for this shoot we needed precise line-up that being able to easily compare the camera's feed and the reference footage gave us which wouldn't have been as dependable from a film camera's video tap. Plus we obviously didn't have the time to wait for developing and scanning to see if we were close enough.

Now if you were looking at whether to buy an HVX or rent a HDCam/Varicam setup for a single shoot the latter would be pretty attractive, especially if the noise is problematic for your application. But even for us, because we are likely to do many such shoots, buying an HVX is still fairly attractive. Though we may well wait 'till NAB to see if Panasonic or a third party does anything to address this issue. Better in-camera noise removal or even something like 'Noise Ninja' as a plugin for Avid or for AE/Digital Fusion that is optimized to dampen the HVX's noise would be great. We already ended up using a noise removal tool in Fusion plus frame-averaging to quiet down the noise for this shoot.
 
Last edited:
Graeme Natress has a very interesting Noise reduction tool in his Filter set one for only $30.00. It supposedly allows you to apply noise reduction only in certain lumanence bands, perfect for the lower level HVX noise. I'll try it and get back to you guys.

http://www.nattress.com/Set1/G_DVNR.htm
 
It's too bad the camera doesn't have a negative gain option. I've used this option on other cameras like the DSR-500 and it cleans up the noise quite well. Although it lowers the camera's sensitivlty, I wouldn't mind giving up a stop to have a cleaner image. (Maybe a firmware upgrade?)
 
Well once it gets dark we are going to try using much more light and use postfx to bring down the levels to create the desired mood. In theory it should work as long as the ranges between levels is the same just brighter. The problem may be in watching hot spots that will try and creep in. The latitude for execution is looking very small.

Also I am going to adjust the core detail to see if it helps. IN the manual it claims it that a lower setting will decrease noise but lose detail. A tradeoff I could live with I suppose. Worth a try. Anyone else tried this?
 
It sounds like these HDV cameras are less sensitive than their DV counterparts, so my guess is they avoided the negative gain option for a reason. With all this talk of banding and noise, and $6K+ cameras with 1/3" chips, and mega expensive storage accessories, the XL2 or 100B are looking pretty good to me...
 
Okay here is the latest on my quest to shoot moody images.

I adjust the detail levels a bit and that took out some of the noise issues. Next we lit the scene with an higher light levels. We stopped down from wide open about 1 stop and got a good high contrast image. Changed the Gamma from D to V it seems to make the blacks a bit darker and more contrasty. Also we reset the black balance as the manual claims it can cause noise if it is off.

On the Dell 24" monitor things looked pretty good with the direct feed. Some noise but not as much as before. The first time we lit pretty flat as it usually give the most room for post adjustments on the DVX. This time we lit the face as hot as possible and adjusted from there. Keeping hard drop offs to the shadows seemed to look good and the darker shadows don't show much noise.

Pulled the footage into FCP and started pushing it around. First thing I noticed was that areas that appeared to be overexposed and blown out were easilly recovered. I had a white hard plastic ball in one shot close to the light source. It was blown out. In FCP I adjusted the levels and was able to recover the highlight. I know I was able to do it because there was a scuff mark on the ball in the highlight. I was able to adjust down and recover the scuff mark and keep the detail in the shadow areas. I was way impressed. Same goes for Faces. I felt I overlit the face in one test but once in post I was able to get the skin texture back and give the shot a nice contrast feel.

I dropped the saturation and pushed the contrast for a nice Bleach By Pass/ Silver retention look and it looked great. The contast totally masked the noise. Even with color corrections I was able to push it pretty far and never see any major artifacting.

Even with subtle crushed blacks the image hides the noise. The CineGamma does have that milky black look that needs to be corrected in post.

The one scene we shot that was pretty flat showed the worst noise problems in the shadow edges. It was blocky and crap looking.

I don't have a HD CRT but the images on my SD CRT and TV both looked incredible. I didn't see the noise at all. I think the Dell is pretty unforgiving. It has a very sharp image.

It is my conclusion that the noise when shooting this way is very acceptable. It was there but not to any degree of discomfort. I don't know about pulling a Matte or Keying I plan to test that later this week. Compositing elements such as 3d images or Matte paintings may be tricky as mimicing the grain dance of the noise would be difficult. Although I could use a noise sample from actual footage to make a transfer layer or channel mask like I had to do back in the day before fancy plug-ins.

I guess the trick is to overexpose a bit if you want high contrast moody images.

I am very happy. I was pretty stressed last night with a feature looming in the near future I was afraid I would have to find a new cam or rent a VariCam. I am confident I can shoot the film I invisioned with sharp detail and low noise using the HVX.
 
holditnowfilms said:
The contast totally masked the noise . . . Even with subtle crushed blacks the image hides the noise . . . The one scene we shot that was pretty flat showed the worst noise problems in the shadow edges. It was blocky and crap looking . . . I guess the trick is to overexpose a bit if you want high contrast moody images.
Click to expand...
Your findings exactly duplicate my own findings with my DVX100! In my Scifest short, I have a very contrasty scene where much of the frame is black, but there are several 100% IRE spikes also in the frame. Guess what? No noise! Then, I have a daylight exterior. Tons of light. I had the ND on it was so bright. But it was OVERCAST, and the exterior was VERY LOW CONTRAST. Guess what? LOTS of noise! So, as JMLang also mentions, it's not merely light level.
 
The HVX is noisy, not overly so but with the blacks not crushed, it can be very obvious. It is a combination of the small CCDs and the DVCproHD codec. On areas of shadowy red it gets pretty nasty. That being said the 100a was about the noisest 3CCD camera on the market, didnt stop people from doing great things.

As far as the XLH, at -3dB it is noiseless... the Z1 is noiseless up to +12db...



ash =o)
 
Well I don't know about you guys, but I'm not too inclined to pay $6K+ for a camera that makes a noisy signal, regardless of 1080i. I was more excited about the DVCPRO50 mode than anything, but after all the attendant hardware and futzing around associated with this camera, you end up spending like $10K and a lot more time than you should massaging the footage.

If I bought this camera and shot a documentary job and then saw banding or blocks in the shadows of the footage, I'd be pretty unhappy about it.

I think Panasonic is getting ahead of themselves with this camera in an attempt to get ahead of the competition. I expect there will be a shload of pissed off filmmakers in a few months trying to filmout from footage shot on the HVX, or project from tape.

I'd rather have a clean SD picture than a noisy HDV picture anyday. Especially in the year 2006/2007 where almost nobody but us video geeks have HD monitors anyway.

I'm sticking with DV25, analog SD, rented SDX-900 (sweet) and film until about the year 2009 when the hardware/cost ratio of HDV will be reasonable and that entry level stuff actually works.
 
Question, has anyone got info on any exposure tests that have been done on the HVX? The noise issue that everyone is refering too sounds a little shaky. Has anyone tried to determine the ASA of the camera, and then tested to see actually how much exposure latitude there is, before we hit noise or grain? Prior to trying to adjust it in post?
 
-1 on the master ped if that's what you are asking. ABB set prior to test.

I will try some messing with the camera levels tomorrow to see if I can pull the same look off I got in post in camera (not the extreme stuff). I am usually reluctant to do too much in cam for fear of getting locked into a look without much wiggle room in post. But some adjustments couldn't hurt I suppose if they kill the noise.

As for shooting a documentary. I would not feel confident doing that unless your subject had plenty of light. Well that or I figure out the problem was a result of settings not the camera. But light level that is about two stops off key starts to fall apart and get noisy. I will bust out the light meter and try to get exacts for those of you who like real numbers and not eyeballing.

In real world doc terms- A room during the day 3pm northern exposure. Full sun outside. One window providing ambient light. Image too noisy on the Dell 24"

Small room. Single 100 watt bulb overhead. 8 ft. Ceiling. Not enough light.

In my doc experience using the DVX I was able to get acceptable footage in those conditions. I shot club DJ's in clubs, back rooms, studios etc. Night time streets under existing lights. Or using a 20w camera light. But I wasn't looking at it on a HD monitor either.

Anyone know the ASA on the DVX?

Please note all these problems could be setting related. I am by no means an expert on this camera yet. Also I would like to see Sony, JVC and Cannon footage on the Dell under simular condiions. Could be the monitor is an issue. Could be I have a tiny brain.
 
1080i24p
Scene file F6 with stock settings
320-400ASA (we had two different meters)

That's all I've tested for now.
Didn't have time for a latitude test.

Plan on doing some more tomorrow.
 
Chris Santucci said:
Well I don't know about you guys, but I'm not too inclined to pay $6K+ for a camera that makes a noisy signal, regardless of 1080i . . . I think Panasonic is getting ahead of themselves with this camera in an attempt to get ahead of the competition. I expect there will be a shload of pissed off filmmakers in a few months trying to filmout from footage shot on the HVX, or project from tape.
Click to expand...
Yeah, but my point is, is that NO ONE can make a $6,000 camera that isn't as clean as a 2/3" camera, costing two or three times as much money! The HVX200 makes incredibly beautiful pictures when properly lit. I just wanted to point out its warts, simply because, since we've all gotten so excited about this new product, I didn't want anyone's expectations sorely deflated only AFTER they've made their purchase, and finally realized what I've been talking about, only after they've ingested their first P2 footage into their NLE!

The first day I got my Sony VX2000, I went out and shot a bunch of beauty shots of L.A. and ran home to ingest. I had a set of ND grads and photographed a beautiful L.A. skyline, and other "postcard" shots I've been wanting to "bank" for stock. Once in my NLE, however, I was sorely disappointed with how soft and noisy my footage was. The same thing happened when I bought my DVX100, only the picture was much prettier, and had way more definition. As I said, when properly exposed, the HVX200 looked every bit as good as a $60,000 2/3" SD camera, only way sharper! That's really something for a $6,000 camera!
 
I think we need to define properly exposed for this discussion. There are a lot of ways a shot can be "properly exposed" depending on the look you are going for and the subject. Properly exposing a grey card is a very specific thing but only helpful in theory.

So what do we mean by properly expose - do we mean avoiding dark shadows? Do we mean avoiding nearly clipped highlights? What latitude should we be working in?

I guess this really comes down to what is the "sweet spot" for the camera?
 
JMLang said:
We tried all the modes, some were so horrible for noise it's hard to understand why they were included at all. The two that showed it the least were Cinelook-D and crush-blacks (I think that's the name).
Click to expand...


I'm con-fuze'd :huh:

I thought Cine-D was listed by Barry as the 2nd noisiest setting.

(hey, can anybody dig up that thread where barry gave an objective list of the settings from quietest to noisiest?)

Are you sure cine-D was the set up?
 
I do also think that the Dell is showing the noise that I see as well. This noise wasn't visible on my old CRT monitor. The footage looked awesome on there, but it looks like dancing distracting noise on the Dell. Has anyone hooked it up to a HDTV? Maybe that would be more softening and give a more pleasing look. For me.. it's somewhat alarming, but it's because I've been shooting SD for so long and noise never really came into play with what I've shot. I have had some footage cross my path that was noisy, but it's usually because of the camera operator error or massive gain on the cam.

How have you been dealing with it in post? Do you think it's a deal breaker? Also, has Panasonic made any offical notice about this noise?
 
Properly exposed grey cards are only safe when shooting film. You can fix overexposure or underexposure in film with little compromise (more grain or increased contrast). When we are talking about a CCD that is sampled through an AD, there is no room for overexposure because the info is lost for ever (digital max) and overexposure looks pretty bad in some cases due to CCD problems. Same applies to underexposure, data lost to black and compression format noise. The HVX produces a blue halo in overexposure, even if it is only 10 pixels that are overexposed. It also produces blockiness due to the compression in underexposure. Vertical bleed is another artifact that is very prominent in HD sensors 2/3inch or smaller. The nonlinear dynamic response of digital cameras with cine settings makes the grey card only a very rough guide.

Accurate comparisons are very hard to make. Changes in post (curves) will also change the results. Every single test I have seen on the net so far in terms of latitude is way out of line, innacurate and far from the truth. Same applies to all noise tests I have seen. And I'm talking about cameras I have used, measured dynamic profiles in different modes, and investigated further in post.

Then we have compression. That also limits dynamic range and adds an extra 20-30dB of noise depending on the scene. All these combined together make it almost impossible to compare, even in the lab and with proper analysis software.

I think the best thing we can do is to judge by size of sensor. An 1/3inch compressed camera will have the latitude and noise of an 1/3inch compressed camera. This is safe at least.

If anyone wants to try a test, just try fit the dynamic range to the format with any controls available while avoiding overexposure (cine settings or ND etc) and the weak low levels of the format by stretching black and similar options. Then, the most important thing, post the results to get similar exposure. Then compare. You shouldn't get huge differences in cameras of same resolution and CCD size. If you do, it's the superiority of the format (less compressed, less noise).
 
Btw, the format that appear noisiest in presets on the camera is technically the format with the greatest S/N ratio and the best latitude if gain setting was the same. This is apparent if you post to produce similar exposure in shadows/midrange/highlights. The noisiest format will be one that pulled the dark info into a better part of the format and actually saved it. You can pull that info down and reduce noise but you can't pull something up out of info that was recorded lower into format noise.
 
You must log in or register to reply here.
Back
Top