View Full Version : HPX3100 forced to work SD recording DV and picture too soft to use.


Pete Bronlund
July 6th, 2011, 04:28 PM
I have Googled as much as possible and found the general consensus is that using an HD camera switched to SD mode will always produce an inferior soft picture to that of a permanent SD unit.

Background: We are using some 30 odd SPX800 and 900 SD (DV_50i) cams to to shoot News for our station under a dedicated P2 workflow with multiple Avid Newscutter editing suites. All footage is recorded DV to make as small as possible file as we have only so much Raid Storage.

Recently an increase in News shooters meant more cameras. Since no maker offers an SD machine any more, we optioned for 3 (for now) HPX3100. To remain compatible with the Avid/P2 workflow the cameras were set to record in SD (downconverted HD 576_50i) and recording DV files. The result are images seriously softer than that produced by the other 30 SPX800 and 900 camcorders and essentially rejected by the Producers. Similar problems arise when feeding SD-SDI into an UpLink LiveEye system. The cameras just look really soft in SD... and adjusting DTL of the downconverted material does not improve things... The HPX cameras just do not resolve enough TV lines when in SD compared to the performance of the SPX cameras shooting a test chart. (it does not appear to be the failure of the downconverter in the camera... feeding the cam's HD-SDI through a full performance Lietch downconverter produces the same magnitude result).

It would appear that this is a problem with Sony XDCAM machines as well from what i have found on he 'net.

Has anyone a suggestion where to next and just why the downcovert looks so soft? The ideal is to increase our Storage and to begin handling HD footage in Editing as well as our SD material however with so many edit suites and an always desperate rush to On Air for News, working with large files over networks is not ideal let alone training up editors to set their Avids into different standards! We need a way to successfully transition to using HD. Interestingly should we ever receive rushes by a Stringer on XDCAM the picture quality is acceptable at SD when added to our system.

Would it be better to shoot in Full HD on the HPX3100 at 1080i-50i (we are PAL based in NZ) record in AVC_Intra 100_50i and use a third party downcovertion program (if there is one... any suggestions?) to create close to DV (maybe even DVCPro) sized files as quickly as possible on a fast separate dedicated computer and transfer them to our main storage? Will the downconverted file look objectionably soft anyway?

In short how are other P2 users meeting a situation like this?

Comments appreciated (any)... thank you.

David Heath
July 6th, 2011, 05:59 PM
Would it be better to shoot in Full HD on the HPX3100 at 1080i-50i (we are PAL based in NZ) record in AVC_Intra 100_50i and use a third party downcovertion program (if there is one... any suggestions?) to create close to DV (maybe even DVCPro) sized files ............
As a very general point, if you wish to shoot HD, but the main current route is SD, there is a lot to be said for originating 720p/50 - not 1080i/25. The latter requires a deinterlace before the downconversion, whereas for 720p/50 there is a discreet full frame available to downconvert to every SD field.

Whatever you do, I'd avoid DVCPro in PAL, and go with DV/DVCAM. They are both very similar, but the big difference is that DVCPRO is 4:1:1 sampling, DV/DVCAM is 4:2:0 for PAL (unlike NTSC). The reasoning behind 4:1:1 was benefits with repeated digital-analogue conversion (when the inputs and outputs of the DVCPRO machines were all analogue), which was important in the early days. Now, it's far better to keep 4:2:0.

Daniel Epstein
July 7th, 2011, 07:24 AM
You may have already tried this but you should try and set the cameras up on charts next to your SD cameras looking at the SD Downconverted output and set the camera detail before it goes to recording or output. This may make the HD look too edgy but the camera should have enough flexibility to make the SD look good enough. It is hard to find a combo setting which looks good in both so I would sacrifice the HD at this point. Also you may be running into the issue of sensitivity of the HD cameras is less than the SD cameras due to the smaller pixels in the HD cameras.
Sorry you can't use my NTSC SDX-900 cameras I have sitting in my office. I would let them go for a song.

Richard Corfield
July 10th, 2011, 01:29 AM
I actually use a Sony EX3 and Final Cut Pro but my method of downrezzing from HD to SD should work for P2 and avid.

I shoot 1080/50i. I ingest the clips onto my macbook pro and place them on a ProRes HQ PAL standard definition timeline. I join the clips up into approx 40minute chunks, then export each chunk. The resulting Prores SD files are used by the editors who are used to ingesting Digital Beta tapes and like having a 40 minute "tape" of footage. The quality of the resulting SD is not bad at all. Not quite up to the standard of Digibeta, but not far off. I've seen my footage intercut with digibeta and you'd be hard pressed to tell the difference.

The key is using best quality Prores. When i first started out with the EX i tried both DV and DVCPro50 downconversions. Neither gave spectacular results, looking decidedly soft in comparison with Digibeta. Prores was a breakthrough for me.

It takes a while to export, almost twice real time. Make sure, when the computer asks you if you want to match the sequence settings to the clips that you are putting on the timeline that you choose "no".

More recently I've attached a Nanoflash unit to the camera which can record standard definition IMX files straight off via the camera's SDI output. You can program the EX to output SD over the SDI. It's always annoyed me that EX3 won't record SD (the newer EX1R can). I convert these IMX files to Prores in the same way. The downconversion time is much quicker, really fast. The resulting standard definition is very slightly better, though you need a magnifying glass to see! Its main improvement is that there are no stepping artefacts on diagonal lines as there are with the FCP downconversion although i must say you really have to look closely at the pictures to see them.

I've experimented with using 720p, but i believe the SD from 1080i is marginally sharper. Not much in it.

That's my experience, works for me.

Christian Magnussen
July 18th, 2011, 10:01 AM
It seems that some camcorders don`t have very good scalers built in and bbc recommendation for several HD camcorders is to shoot hd and use a good scaler instead.But some are better than others, or at least PDW700 in SD mode looked good enough for NRK in Norway.

The best option would be to use a good scaler such as AmberFin (http://www.amberfin.com) which use the same methods for scaling as Snells Alchemist. But with news you usually don`t have time to go trough that process...

When I have to deliver sd I shoot sd with my hpx500, so far no complains.

Tom Hardwick
August 14th, 2011, 07:26 AM
The problem is, Pete, that HD cameras always shoot HD (1920 x 1080) and if you switch them into their SD mode they do the downconversion to 720 x 576 on the fly between chips and card. You and I know that to do this downconversion properly takes time and space (TMPG or Vdub) and lots of gigs. If you want it done properly it takes lots of money (Snells Alchemist).

So you can bet that downconverters built into camcorders do the cheapest, just acceptable (maybe) job. And the results end up looking soft, with artifacts and Moiré by the bucket load.

tom.