disadvantages of fx1
 

Hi
I would like to buy the sony fx1 camcorder. I read somewhere that has few disadvantages like autofocus problem and noise recording. I cant spend over 3500$ to buy other HDV camcorders. Now i have the sony trv70 which is normal dv camcorder which has problems too like autofocus. 3500$ is too much for me so i want to make sure that i will not have such problems with fx1.
Is it true the autofocus problem on fx1?
thanks

I don't use autofocus, so I can't address that. I never noticed any noise in the recording, except for the HD10 from JVC, because it's a 1-ccd camera.

heath

not video noise but audio noise

Didn't have that, either, including using the camera mic and hooking up an external via an XLR and 1/4 inch (I believe, or 1/8) adaptor.

heath

yes there are occasions where the auto focus is hunting and seems to take a long time to figure it out. but it's never been an issue or problem for me. I've done a lot of run-and-gun documentary stuff recently with it with very few issues.

audio sounds great. you will pick up some motor noise, but it's very, very low, but it's there.

To me, the biggest disadvantage of the camera is none of the above, but the HDV format and MPEG-2 compression. But it's what I got, I work with it, and love it. If I was unhappy with it, I would've sent it back long ago, but the pictures still blow me away.

Mike, what's your workflow? Do you use Aspect HD/PPro 1.5.1 to edit the footage?

My current workflow has been DV editing on my laptop. I've been on the road for the last 3 months- just got home yesterday and leave today for another week- so I haven't really setup anything for HDV editing as of yet. 3 months ago, I wasn't really impressed with any of the HDV editing options.

My future workflow will be editing directly off a P2 card in my G5 or G6 using FCP HD. <wink>

Yes... I'm tempted by G5/FCP 5 too... it's just way off my usual Wintel-based path... I guess I'll stick with the PPro for now. Now, if only HD Link (Cineform's capture/converter utility) didn't glitch on capture, it'd be the day.

Alex,
I think the glitch is hardware based. I had a similar problem that was diagnosed as Rambuss 800 being too slow. If you are using the purchased plug-in, my work around was to import m2t files from camera then convert the imported files...it makes a bit more work, but is less processor intensive...the results were very nice.

Sean, well, the whole purpose of Aspect Hd to me is *not* to have to upgrade my hardware...

Workaround: Yes, I also capture m2t first with CapDVHS, then I fix GOP errors with Womble's tool, and only then do I Convert it using HDLink. It's OK, but of course I'd expect HDLink to do all that in one pass as it should...

Hey all,

We're getting into editing territory. We have an excellent page on HDV editing here:

http://www.dvinfo.net/conf/forumdisplay.php?f=63

Thanks,

heath

Compression and noise
 

Heath and Mike: Funny how we all get different impressions of stuff. I've never seen any compression artifacts or image degradation at all from the HDV MPEG compression. As for image noise, it's there for sure if you look close enough. I've yet not seen a single segment without noise. But it's at an acceptable level i think.

the image noise you're seeing is most likely from the MPEG compression. Most artifacts aren't noticable, you're right, but they're there.. you especially notice them when you effect or treat the video.

here are some screen grabs clearly showing artifacts. again, when played back on a 1080i monitor, they're very hard to see, and the image, YES EVEN FAST MOVING IMAGES, look fantastic.

http://www.miketiffee.com/hdv/sugarbowl1.jpg
http://www.miketiffee.com/hdv/sugarbowl2.jpg
http://www.miketiffee.com/hdv/sugarbowl3.jpg
http://www.miketiffee.com/hdv/sugarbowl4.jpg

Awesome shots.

Being the incredible Auburn fan that I am - you wouldn't have any short clips of that game I could download do you?

It would be awesome to see some of it in hd(v).

Mike, I dont think I mistake the mpeg compression artifacts for noise. Noise levels for the FX1 are so low that for zero gain shots of moving parts, I cant even see any in magnifications of my clips. But it is still there, and the evidence is to look for a stable shot of a big flat surface. When I play the clip the surface should look like a still image. The more clearly I can detect that it is a video as opposed to a still image, the more noise it is. If they look exactly the same = no visible noise. Such parts of a clip will not suffer from bad compression artifacts. For the price the FX1 is a magnificent piece of work, and dont mistake me for a critic of it. The noise levels are fully acceptable.

Thanks for the example images. I clearly see the blocks of compression artifacts. Slightly disappointing, but as you say probably not even noticable when played in motion. Perhaps the shots overloaded the maximum DVI bandwidth, and as a result the camera couldnt cope.

Hi Mike, well those images impress me. Are they deinterlaced. They look a lot sharper than I achieve shooting less action (for example table-tennis) - but that is not surprising since I have much to learn with using the camera. How did you do it.

Andrew

125 shutter speed- and yes, I believe I deinterlaced the stills in photoshop.

Alex,

Let's go back to what you said about GOP errors. I don't get any. Ever.

So, let's take a look for the root cause. If HDLink gives you problems when capturing, and CapDVHS gives you problems (it must or you would not be correcting GOP errors, right?) then it isn't the software package. Have you tried directly into a HDV project in Premiere Pro 1.5.1 ?

It is probably either your PC firewire card, hard drive, memory, or worse yet, your camera. Can you find a different PC to capture to?

What do you mean by "then I fix GOP errors with Womble's tool"? What are you seeing? This bothers me since it doesn't happen to me (knock on wood) and I am a big Cineform fan.

Yeah, I have to say that in my 30+ hours of footage with the FX1, I've had one dropout and no "GOP errors," whatever those might be.

I've found the color does have a little trouble keeping up with heavy manipulation, but nobody I've worked with seems to notice in the end product; it's just me in the editing room that sees the diff. And usually when it's noticeable to me it's because I f***ed up the lighting somehow and am trying to make something out of nothing. Light decently well and this won't be a problem. :)

PPro 1.5.1 has worked very well for me, but I'm running 2gig RAM and a 500GB RAID 0. You ask me, the extra $1K to build a box like that is totally worth it.

GOP errors: I'm out of my depth here. It seems like Womble's GOP Fixer tool finds "timecode GOP errors", whatever they might be.

I highly doubt that my hardware (PC/camera) is actually faulty. I do think that somewhere there's not enough processing power. I also think that Aspect HD should precisely enable "normal" PCs to process HDV. See my PC's specs below - they meet Aspect HD's requirements. Yet something is not working right - see below.

CapDVHS also seems to work just fine.

Workflow that, umm, works: Capture m2t from FX1 to PC via firewire with CapDVHS. Once captured, fix GOP errors with the Womble tool (I *think* that if I omit this step, Aspect HD conversion of the long segments may or may not stop prematurely.) Once fixed, convert that m2t to Aspect HD, Medium quality, Progressive. (Note that my taped footage is Cineframe 30, which is simply a high-quality in-cam deinterlacer so the output is de-facto 30p.)

Workflow that does not work: Capture in PPro 1.5.1's cap utility. Problem: digital drop-outs randomly occur.

Workflow that does not work: HDLink's standalone capture. Problem: digital drop-outs randomly occur, whether I choose to keep m2t or not.

My PC is Win XP Pro, P4 3.0 Ghz HT w/512K on-board cache, Memory Kingston 512Mbx4=2Gb timing 2-3-3-8, dual channel DDR-SDRAM PC3500 in "performance mode" set in BIOS, on mobo Asus P4C800-E Deluxe with on-board firewire, 400 FSB. The PC is *not* overclocked. Capture in all cases happens onto a HUGE SYSTEMS Raid 3@2K, via SCSI PCI 32 card. Actual (conservatively measured) sustained bandwidth is 65Mbs read, 55Mbs write.

You say that CapDVHS works fine, but then you say you have to correct errors. Which is it?

And I see where you say "whether I choose to keep m2t or not", but have you tried just capturing the M2T with no conversion?

By the way, your PC is about the same as mine and I have no problems.

Steven, CapDVHS is fine in terms of it actually captures m2t's without any visible problems. Womble still finds and fixes errors in m2t's. Check your m2t's and see whether Womble finds GOP errors in them - it'd be interesting to see whether that error report is even relevant.

I downloaded and installed Womble. Interesting program, by the way.

I captured a M2T with HDLink - just a little over 10 seconds.

When I run the MPEG Video GOP fixer, I use the test option of Scan GOP time code, video and audio PTS and GOP size errors (Read-ONLY)

But I do not understand the response. I looked in the help, but there are no examples.

Here is what I get - but what is the [20] ?

http://www.stevengotz.com/support/womble.jpg

In my version of Womble, it clearly shows GOP TIME CODE as [1]

...and then it shows errors [1] = 2356 (for example), and fixes them.

So it's FOP Time Code errors, in my case.

No other errors are ever found.

I do not know what [20] refers to on your photo snap, sorry.

http://www.uploadhouse.com/images/406601443sugar_0.jpeg
http://www.uploadhouse.com/images/939282290sugar_1.jpeg
http://www.uploadhouse.com/images/544597664sugar_2.jpeg
http://www.uploadhouse.com/images/872279471sugar_3.jpeg

I couldn't pass up getting rid of that blue haze that the FX1 gives.

Alex,

I highly suggest you find a way to test your camera on a known good PC.

Installed HDLink on an ancient PC with AMD 1.8Ghz, 512 Mb memory.

Captured about 2 min. of footage using HDLink with on-the-fly conversion into aspect HD codec while keeping m2t file.

Guess what, no drop-outs whatsoever (sure it took 3x real time for the AHD conversion, which was expected with the slow processor.)

So maybe there *was* something in my "fast" PC that screwed the whole thing up. I'm now capturing/transcoding 1 full tape of HDV footage using HDLink on the ancient PC (see above) to see if glitch-free operation was just a coincidence or what. Will let you know later.

I often get shouted down on this one, but the problem could be with the RAID. I do not know why, but many people have trouble capturing to a RAID.

My old PC is capturing on the RAID 0 - 3ware Escalade card sits in the PCI 32 slot and feeds 4 WD HDDs.

My new (bad?) PC has SCSI card in PCI 32 slot that feeds Huge Systems' Raid3 formatted with 2K sectors.

So both are capturing to the RAIDs, just different ones.

If I were you I would throw a cheap IDE drive into the system and capture to it. Just to see what happens.

Steven, you were right all along.

My "new" PC seems to be having problems feeding HDV video files off its RAID system. Weird, since I'm getting at least 65Mb/s sustained data read test results.

When I moved the whole project and its files off the RAID and onto another regular IDE drive, tada - PPro no longer quits, nor does it produce any errors. Go figure. I guess something isn't kosher in my "new, fast" PC's SCSI/RAID setup.

I now think that my problems with CAPTURE via HDLink on new PC were of the same exact nature. I don't have time to test-capture on IDE drive, but I'm 80% positive this will work fine.

So it's not a memory/processor problem per se, after all.

Thanks to Steven and everyone who helped!

Glad to help steer you in the right direction. I just want everyone to have as much fun with HDV as I am having.

What I don't like with my FX1 - The Viewfinder seems to show a sligthly smaller picture than the final record which can be a nuisance.

Richard

#1 issue with a misbahaving scsi raid is shared resources/PCI bandwidth. If you have a 64bit raid card in a 32bit slot, or your graphic's card/capture card is sharing resources with the array, etc.,etc.

Most folks run a disk speed test, getting good results, not realizing that the raid appears to be correct because it's the only thing running at the time of the test, as the test is done internally. Such a test only indicates that the raid is communicating properly with the disks. What you are experiencing is how the raid is working when other hardware/software is also hogging resources.

Pete

Richard,

Use a monitor, CRT (no flat panel), and even if it isn't HD, NTSC SD is better than nothing.

heath

I have to admit that I am switching to manual focus more often with my fx1 than I do with my vx2000. But nothing compares to the autofocus problems I have encountered with the XL1's I have used - they are really touchy. Overall it is fine, but low light does create focus problems quickly and the fx1 is not good in low light compared to the vx2000 and this could be a problem for some event or wedding situations that will not allow additional lighting.

No capture problems for me yet in Vegas or Premiere.
I am worried about the 384K compressed orginal audio however, and it seems ashame that my vx2000 records better audio than the fx1.

I shot two one hour shows twice with two cameras my FX-1 and a rented ZU-1, since it was a one man gig, I had to use autofocus. Much to my surprise the autofocus did a fantastic job - it misses occasionally, but it´s very reliable indeed.
In 8 hrs of footage, I estimate 4 minutes was out of focus, no problem working around it in post.

Alex, you seem to have found the problem, but when you listed your specs you stated 400fsb. Did you mean to say 800fsb? I am asking because I had a friend running his 3gig p4 at 1.5gig for months and never noticed.