Problem with HDV: artifacting in the blacks?
 

Hi

there was a HD talk held here last night (with HDV) and one problem that an HD expert cited when he toyed around with footage from the FX1 that he got and uncompressed is that he got major artifacting in the already crushed blacks.

He would see black blocks appear here and there and there was nothing he could do to fix it. So this problem may carry down the chain until broadcast.

This is scary for me, if I want to shoot anything for broadcast, if post will turn out to be expensive as I try to fix such problems.

Does anyone else see this issue?

I would just like to add that this is very possible with an MPEG-2
compression system (you can sometimes see it on badly encoded
DVD's as well). And this would be hard to fix in post indeed.

I don't have a HDV camera myself so can't attest to wether this
"issue" exists, but it certainly is possible.

I have had my FX1 for just over a week and have only shot about 1 hour of HDV so far. So my comments are based on this limited experience. However I suggest they are more valid that someone who does not have the camera and has mysteriously got hold of some video file from someone in some form and has then MESSED with it and passed some judgement. In the 1 hour of video I have seen no artifacting of ANY type, all shot on normal Sony Premium. Shots are both indoor in very dark room ( gain was at 18db, f2.2) to snow with ND2, 0db, f8. I am thrilled with the camera. I have seen comments that the FX1 is not as good as either the VX2100 or DVX100 in low light. That is probably the case. What I can say is that in DV mode it displays less video noise than the DVX100 even at 18db. I rent a DVX100 and AG DVC200 to shoot in the theatre so I have a lot of video in what amounts to very similar lighting condition. The FX1 last night 0db, F3.4 was about the same brightness as the DVX100 at 0 db, F4. They weren't used at the same time its just that the lighting in the theatre is pretty consistent. However on the DVX100 video the dark areas usually display some noise wereas the FX1 has absolutely none. Before the show I shot some HDV, this was just working stage lights ( much darker ) boosted gain to 9db, F2.2, image is beautiful, no noise, sparkles compared to DV.
The performance was shot in DV with the FX1, TRV50 and HC1000. As one would expect the HC1000 and TRV50 were very similar in colour and noise levels but the FX1 colour seemed richer, sequines on dancers costumes stood out in detail, very impressive. I certainly will not be renting the AG DVC200 in the future though may rent the DVX100 as second camera to the FX1. The AG DVC200 is a very nice camera and does much better in low light but with the FX1 performing in this range of parameters I have no need for lower light capability than I need. At 0db and F3.1 I have enough DoF on stage not to worry about focusing issues. From were I shot the show I was 17M from front of stage and 23M from the back of stage. Focus was set at 19M for the whole performance providing ample D0F across the whole stage.


Ron Evans

<<<-- Originally posted by Ron Evans : The performance was shot in DV with the FX1, TRV50 and HC1000. -->>>

Did you notice how the exposure settings compared between the HC-1000 and FX-1? We've been wondering how the PDX-10 (same lens and CCD's as the HC-1000) would compare to the FX-1 in another thread....

Hi,

Yes, the HD expert would have messed with it for sure, from everything from trying to uncompress it, to trying to use it in regular HD applications, because in our industry, there is a lot of post production that is anticipated to be done on the video (from effects to CGI) and it's important to make sure it holds up.

We're not just shooting and making simple cuts.

Boyd, a colleage shot with the HC1000 and I will check with him tomorrow for the settings but the TRV50 is a 1ccd with the same CCD as the HC1000 and 14bit DXP. Data code from the TRV50 shows 0db, F2.8, 60th and from the FX1 0db, F3.4 at 60th for similar picture brightness, same scene, at least by my eye!! The HC1000 was close to stage at one side so may have different settings. The TRV50 was on a tripod right next to the FX1.
Ronald. My reference was to the fact that the method of capturing files from the FX1 is important in the evaluation of performance. Do you know how this was done by this "expert". Were they captured by some "hacked " method then further analysed by what method? For me to believe I would like to know :"exactly what was used to capture - software/hardware and how the file was then analysed. That way I can pass judgement on the expert. Dealing with new technology is not straight forward and it is easy to pass judgement based on ignorant analysis, even by an expert. Blocking usually occures by a codec having problems encoding or decoding data. This can occur during the encode or the decode. As an example, my old Hitachi DVD player often has problems with burned DVDs either skipping or displaying blocks---IS this a problem of the DVD and MPEG2 encoding ?> No it is a problem of the player. Do I make a judgement about DVD-R's based on this player? NO Just has everyone has had to learn to burn DVD's on a very clean PC, no interupts etc, we will have the same issues in dealing with creating clean MPEG2 files during editing ( it is the same issues as DVD burning MPEG2). MY read on the blocking/artifact is that it was created on the PC doing the capture and analysis and is more characteristic of a playback problem. ie the PC wasn't up to the task.

Ron Evans

Boyd, just checked for the HC1000 parameters, they were f1.6, 0db, 60th. We haven't checked to view side by side so images may be slightly different in exposure. The HC1000 intent was to get close ups. We will try and do some more tests over the Christmas break comparing alll the cameras at our disposal ( VX3 Hi8, TRV730 and TRV740 D8's, PC10, TRV50 and HC1000) However a non scientific assessement would be that for DV the DVX100 is at least 2 stops better than FX1 though picture is not as clean, TRV50 is maybe 3 stops worse than FX1 and HC1000 is a stop or two worse again. THe 14bitDXP in both the FX1 and TRV50 does not seem to give as broad a latitude as the PC10. I have no scientific way of measuring this just my eye says I see a lot more shadow detail from the PC10 than either of the FX1 or TRV50. Since the PC10 is getting a little old now it suffers from more colour noise than the newer cameras.

Ron Evans

Ron, these two statements seem to contradict each other. The first indicates a 1/2 stop difference (if the FX-1 was shooting at f4.0 that would be one stop):

the TRV50 shows 0db, F2.8, 60th and from the FX1 0db, F3.4 at 60th for similar picture brightness, same scene

and here you say 3 stops:

TRV50 is maybe 3 stops worse than FX1 and HC1000 is a stop or two worse again

If you look at this thread, John feels that the FX-1 is two stops slower than the VX-2000. My own tests showed that the PDX-10 (same CCD's and lens as the HC-1000) was about 2.5 stops slower than the VX-2000. That would make me expect to see maybe a 1 stop difference between the HC-1000 and FX-1, not four or five stops. If the FX-1 were shooting at f3.4 0dB and the HC-1000 was at f2.8 +18dB that would only be a 3.5 f-stop difference. Perhaps were using the term "f-stop" differently?

Hi Boyd non of the comparisons I have done are at all scientific in their measurements. I have just rechecked the data codes on the tapes for the three events I have shot with the FX1 and the TRV50.
The first was very dark school dance
FX1 F3.4, 9db 60th
TRV50 F1.6 6db 60th

Ballet school concert

FX1 F4.0, 0db, 60th
TRV50 F2.8, 0db, 60th

The HC1000 at the same ballet concert in a different location

HC1000 F1.6, 0db, 60th

In the next few days my friend and I will see if we can get a little more accurate comparison. But our feeling is the HC1000 is not as sensitive as the TRV50 which is not as sensitive as the FX1. With a stop or two between each of them. I have tapes shot with the DVX100 in the same theatre in similar lighting and will look these up and check in relation to the TRV50 later today. Until then the only comparisons I have are as above. During the performance the TRV50 will stay the same but I adjust the FX1. Until I edit the show I will not know how true the differences are between them.

Ron Evans

Hi Boyd yet another update!!!
Looked up a show this summer where I used rented DVX100, and my TRV50 and PC10.
Data codes show:-
DVX100 F4.8, 0db, 60th
TRV50 F1.6, 3db, 60th
PC10 F1.6, 0db, 60th

Using the PC10 as a reference, the FX1, under similar condition would be F2.8 to F3.1 at 0db( if I have done my ratioing correctly!!) or a lot less sensitive than the DVX100. This would support Barry's findings re FX1 sensitivity compared to XL2 and VX2000. Bear in mind my finding are with my eyes NOT by any measurments.

PC10 and DVX100 were side by side and the TRV50 was used at an angle so was not seeing as much reflected light. Again my placing the FX1 somewhere between the PC10 and the DVX100, the TRV50 is not quite as good as the PC10 and I expect the HC1000 is not as good as the TRV50 ( has to suffer the light loss through the prism block for the 3 CCD of the same type and the same 14bitDXP). Latitude of the PC10 was not far off the DVX100 but TRV50 is not as good as either PC10 or of course the DVX100. Video noise level is lower on the TRV50 than either the DVX100 or the PC10. I guess Sony chose to improve noise level and sacrifice the latitude?? I will do some more tests but suspect the PC10 latitude may be better than the FX1!!. ( An interesting note is that I expect that in present day dollars the PC10 might have cost more than the FX1!!!!) I know that when I got the TRV50 I was quite disappointed in the performance compared to the PC10 ( the image clarity and colour are better on the TRV50 than the PC10 but the increased latitude I was expecting from the 14bit DXP Sony advertising was infact poorer. I think the 14bit DXP shares out the available bandwidth giving preference to the mid and high ends of the spectrum hence the nice colour and detail but also loss in shadow detail. Not quite what I wanted or as advertised by Sony) I obviously need a black stretch control on the FX1!!! An interesting characteristic of the 14bit DXP is that even in manual on the TRV50 it changes how it distributes bandwith. If for example exposure is set at a medium shot of the stage showing nice character detail and clearly exposing folds in the side and back curtains or costumes, pulling back to a full stage view will result in all the black detail disappearing!!! Zoom in again and it all reappears!!! This characteristic is not as bad on the FX1 but I think it is still there. On the PC10 this does not happen. What you see is what you get at any zoom position.

Ron Evans

Clarification (?)!
 

Ronald,
Please let me clarify a few things...
The person who mentioned this "problem" with HDV didn't happen to be on the panel at the Documentary Organization of Canada? (DOC) on Tuesday?
Well I attended the meeting and I presented an FX1 to demontrate and discuss....we'll let's just say the panel member who was trashing on the camera also stated that HDCAM is "done" as a format and is not high enough quality for production requirements. Fair enough....he is a well regarded HD consultant/post production supervisor for big budget projects.
That's fine for the high-end feature market, but come on...the last camera that this panel member used was admittedly $417,000 CDN fully suited up (with glass)!
So certainly, when he is used to shooting and/or color correcting in 4:4:4 on a Davinci the HDV of course won't come close....to me that's obvious.
Most of the people attending the meeting (and future HDV users) shoot their projects on PD150/170s or DVX100s...so far I have been more than impressed with HDV and it's relatively high compression VS DV (Viewing HD component on a 17" WS Sony Luma Professional monitor).
Personally, I'm getting a little tired of people bashing of this camera...I wonder how many out there have actually USED the camera and viewed it on a professional HD monitor. Sure it's not perfect, CF24 sucks and the post end needs more support but CF25 with the Z1 will answer that!
I hope these forums will "progress" into how to get the most out of the camera throughout the production chain...not the same old my-gear-is-better-than-yours discussion every time a new model comes out :-)
FWIW my 2 cents,

Brian Broz

Hi Brian!

Welcome to the board (28 posts, you're an old timer practically). Yes, it was the guy from DOC and yes, he had a major bias towards higher end gear. He told me afterwards that he didn't like the XL2 or DVX100 either, and just told me to test the heck out of all the cameras I am considering. Good advice, but for a guy like me who is planning to buy another camera system next year, I want to make sure I am making the best choice for my needs.

At anyrate I'm waiting for two things:

1) the Z1U to come up and people to test it.
2) to see how the Sony HDV holds up in film transfers.

and maybe 3) to see how good the JVC 24p HDV holds up in film transfers.

I hope people do not think I am trashing the FX1---I love it. It produces beautiful video in both DV and HDV. I am not a professional, just a retired guy who has a very expensive hobby!!! I shoot dance and theatre maybe once or twice a month mainly for archiving for the companies using my TRV50 and PC10. If I am asked to do something a little more important rent DVX100 and AG DVC200. I still have my VX3 Hi8 that I used until it just became uncompetitive with the 1CCD DV cameras. I didn't buy the VX1000 or VX2000 because for me they lacked the controls of the VX3 that have now been returned with extras in the FX1. True it may not have the low light performance of the VX2000 etc but in the environments where I shoot it has ample headroom and that is all I care about. And in HDV it just blows away all the cameras I have experience with, absolutely no comparison. Is it perfect , probably not but what does one expect for $5000Can. Seven years ago I paid $3600 for my PC10 and I think it was 10 or 12 years ago $6000 for my VX3. In comparison the FX1 is incredible value. It might cost more to edit and watch the video!!!!!

Ron Evans