HVXitis-Deadly or just contagious?

LighthouseMedia

LighthouseMedia

Well-known member
Yes I'm officially diagnosed with stage 3 advanced HVXitis. Anyone else catching it? The more I shoot with this camera the more addicted I get. In fact in its messing up my editing because when I go back to work with older footage from the DVX and DV500 etc, I get depressed, my color correction doesn't hold up :( my footage looks flat etc... I'm simply loving this camera. Its far from perfect but work with and around the range of limitation and the footage is simply beautiful. Thanks Panasonic for giving me a tool that is making we want to get out and shoot again! I fear the only cure for this disease maybe a little "RED" pill. LOL aside from that I fear I'm a lost cause.
 
DITTO.. brother. I think a 12 step program is in order to cope with the depencency and addiction.


I also had the MAJOR displeasure of editing some HDV footage this past week. Man that is a system nightmare. DVCPRO HD is nothing for my G5 to burn though.. but something about the HDV made the processor rev up... graphics card sputter and chip.. and just generally play all jittery and squirelly... Not IMPRESSED on the post side, but I guess that is why it's best to convert it to DVCPRO or another format for editing.

Either way... we are in a good spot and I too am enjoying the shooting. I shoot for people for FREE again and that's been a while for me.. Just enjoying the process!
 
Justyn said:
I also had the MAJOR displeasure of editing some HDV footage this past week. Man that is a system nightmare. DVCPRO HD is nothing for my G5 to burn though.. but something about the HDV made the processor rev up... graphics card sputter and chip.. and just generally play all jittery and squirelly... Not IMPRESSED on the post side
Click to expand...

As I understand it, HDV is so compressed that the computer has to really work to run the codec (compression/decompression). I played with some native HDV in FCP on a Dual 2.7 G5, and it crawled.

I'm editing DVCProHD 720/24n on my four-year-old Dual 1.25 G4, and I never get so much as a hiccup. Pretty cool.

By the way, the HVX is not addictive. It's terminal -- at least where my wallet is concerned. I've bought the Marshall, M2, lenses, bigger tripod head, batteries, etc... since getting mine. Financially, I'm having a near-death experience.
 
That is why any HDV footage received is immediately converted to DVCPRO HD. But I have not received any (decent) HDV footage yet! It's all been HVX baby!
My webmaster will be posting this week some great footage. I have more in the cue and packages sitting on my editbay.
 
darwinandpaine said:
As I understand it, HDV is so compressed that the computer has to really work to run the codec (compression/decompression). I played with some native HDV in FCP on a Dual 2.7 G5, and it crawled.
Click to expand...

It's because HDV isn't compressed frame-by-frame, but rather it's compressed into long groups of frames. So to decompress a particular frame, the computer can't just decompress that particular frame (!) Instead, it has to decompress the I-frame (the first frame in the group), and then various frames between the I-frame and the target frame, applying the encoded changes to make sure that it's updating the frame to an accurate representation, so that the next frame's encoded changes will work properly...

To grossly oversimplify it, think of a group of 15 frames, each one encoding only the changes. So let's say the frames look like this:
WOOD
WORD
WORE
WIRE
WARE
WART
DART
DARE
DATE
LATE
LUTE
LUKE
PUKE
PIKE
BIKE

When encoded, let's do a simple compression system consisting of two characters: the position, and the new letter. So in WOOD, to convert that to WORD, we change the third letter to "R", leaving the rest alone. That could be stored in shorthand as "3R" (3=third letter, R= letter to change it to).

So the compressed data stream would look like this:
WOOD
3R
4E
2I
2A
4T
1D
4E
3T
1L
2U
3K
1P
2I
1B

That's quite a bit more compressed, our "compressed" system takes up about half as much space as if we'd stored the full words, right?

In those 15 instances, there's only one letter changed per "frame", yet the first frame is "WOOD" and the last frame is "BIKE". So HDV would have to uncompress the first frame ("WOOD") to get the starting point, and then uncompress the second frame (to turn WOOD into WORD), and then uncompress the third frame (to turn WORD into WORE) and on and on and on, updating the changes between the frames, until it finally gets to the one you want ("BIKE").

(in actual HDV it's not quite that bad, because there are bidirectional frames and such in the HDV GOP, but in reality it can mean having to uncompress up to 7 frames to get to the one you want).

Contrast that with DVCPRO-HD, where every frame is compressed individually. If you wanted frame 15, the editor would go uncompress frame 15. Instantly. No waiting.

So HDV can be bandwidth-efficient, but it's a real pig in terms of processor speed and computer performance.
 
Barry... how do you keep all of that knowledge in your head? I think I'd have to throw out all the useless jingles and voices in my head to get enough room to fit a fraction of what you grasp... and that's why you rock. You help out the uninformed and the misinformed like me...


I can certainly attest to the HDV sucking down my system. I was horrified and quite amazed... It made the fan kick on the G5 and processor whirl.. and the graphics card/monitor started chirping... What the hell is that good for? I don't think you could edit this HDV on a G4 at all.... and I edited varicam footage on a 1 ghz powerbook with only 512 of ram on it a couple of years ago... and the HVX footage I'm sure could be done on the same or even something slower....

Why in these comparisions do they not mention that.. and how you need a badass comp to handle this mess.....

Thank P2... In P2 I trust!
 
Barry_Green said:
WOOD
WORD
WORE
WIRE
WARE
WART
DART
DARE
DATE
LATE
LUTE
LUKE
PUKE
PIKE
BIKE

When encoded, let's do a simple compression system consisting of two characters: the position, and the new letter. So in WOOD, to convert that to WORD, we change the third letter to "R", leaving the rest alone. That could be stored in shorthand as "3R" (3=third letter, R= letter to change it to).

So the compressed data stream would look like this:
WOOD
3R
4E
2I
2A
4T
1D
4E
3T
1L
2U
3K
1P
2I
1B
Click to expand...
That must be THE most understandable explanation of inter-frame compression on Earth! Barry scores again! Did you actually come up with this analogy?
 
Glad it helped. That was just top-of-my-head stuff, I hoped it would be clear.

It also helps illustrate why a dropout in HDV is so nasty. Since every frame is derived from the previous frame (not literally, but in our oversimplified explanation), think about it -- if one little bit gets corrupted, it'll corrupt everything else that comes after it. That's why any little dropout affects an entire group of frames; on the Sony and Canon that means a half-second of corruption; on the JVC it means 1/4 to 1/5 of a second.

That same dropout on DV/DVCPRO50/DVCPRO-HD would only affect the current frame. And can usually be repaired in post. But in HDV, on the Sony system it just aborts the entire GOP (group of pictures) because any corruption propagates throughout the entire group. JVC just lets it roll, so the corruption results in banding errors; Sony just freezes. Don't know what Canon HDV does on a dropout.

GOP compression is way more efficient in terms of bandwidth, but what one hand gives, the other takes away: it's more susceptible to dropout corruption and it's way more processor intensive.
 
Barry you are giving me a brain inferiority complex. Thanks for the lesson. I've heard it stated the hdv takes 7 times more processor power. I'm just curious though, how are all the people using the hd100 editing? In real world time, how much more time consuming is it working with hdv? Just wondering, thanks
 
Barry.. It did help alot and thank GOD for the HVX. I don't think I could handle editing HDV natively.. It really sucked!
 
Scrappy said:
I'm just curious though, how are all the people using the hd100 editing?
Click to expand...
Many HDV editors are choosing to transcode to a different codec when editing. Something like CineForm, or DVCPRO-HD, or Apple Intermediate Codec, or Lumiere, or CanopusHQ... there are lots of "editing codecs" out there, but it means transcoding (uncompressing and recompressing in the new codec) before you can do that.
 
You came up with that off the top of your head????!!!??!?!?!?! Check out the big brain on Barry! Does anyone else feel like a complete idiot? Or, is it just me a Scrappy?
 
Barry_Green said:
It's because HDV isn't compressed frame-by-frame, but rather it's compressed into long groups of frames. So to decompress a particular frame, the computer can't just decompress that particular frame (!) Instead, it has to decompress the I-frame (the first frame in the group), and then various frames between the I-frame and the target frame, applying the encoded changes to make sure that it's updating the frame to an accurate representation, so that the next frame's encoded changes will work properly...

To grossly oversimplify it, think of a group of 15 frames, each one encoding only the changes. So let's say the frames look like this:
WOOD
WORD
WORE
WIRE
WARE
WART
DART
DARE
DATE
LATE
LUTE
LUKE
PUKE
PIKE
BIKE

When encoded, let's do a simple compression system consisting of two characters: the position, and the new letter. So in WOOD, to convert that to WORD, we change the third letter to "R", leaving the rest alone. That could be stored in shorthand as "3R" (3=third letter, R= letter to change it to).

So the compressed data stream would look like this:
WOOD
3R
4E
2I
2A
4T
1D
4E
3T
1L
2U
3K
1P
2I
1B

That's quite a bit more compressed, our "compressed" system takes up about half as much space as if we'd stored the full words, right?

In those 15 instances, there's only one letter changed per "frame", yet the first frame is "WOOD" and the last frame is "BIKE". So HDV would have to uncompress the first frame ("WOOD") to get the starting point, and then uncompress the second frame (to turn WOOD into WORD), and then uncompress the third frame (to turn WORD into WORE) and on and on and on, updating the changes between the frames, until it finally gets to the one you want ("BIKE").

(in actual HDV it's not quite that bad, because there are bidirectional frames and such in the HDV GOP, but in reality it can mean having to uncompress up to 7 frames to get to the one you want).

Contrast that with DVCPRO-HD, where every frame is compressed individually. If you wanted frame 15, the editor would go uncompress frame 15. Instantly. No waiting.

So HDV can be bandwidth-efficient, but it's a real pig in terms of processor speed and computer performance.
Click to expand...



This is the kind of thing you would post on your wall, and silently never say anything about, knowing your friends, or others who use hdv, will one day see, and read, then in the back of their heads, they silently walk around with the knowledge of knowing that their product is inferior to yours, and it all makes sense to them now.............
 
Barry_Green said:
It's because HDV isn't compressed frame-by-frame, but rather it's compressed into long groups of frames. So to decompress a particular frame, the computer can't just decompress that particular frame (!) Instead, it has to decompress the I-frame (the first frame in the group), and then various frames between the I-frame and the target frame, applying the encoded changes to make sure that it's updating the frame to an accurate representation, so that the next frame's encoded changes will work properly...

To grossly oversimplify it, think of a group of 15 frames, each one encoding only the changes. So let's say the frames look like this:
WOOD
WORD
WORE
WIRE

(snipped)

.
Click to expand...

I'm an owner of an HDV camera and I learn more about my camera and format from Barry than anyone else on the web.

There are people out there who may know more than Barry (probably not many though) but I've yet to come across anyone who so frequently and willingly answers questions and most importantly, does so in a way most people can understand.

Barry Green is a legitimate reason to buy the HVX. The freakin' Oracle of Las Vegas.
 
Scrappy said:
Barry you are giving me a brain inferiority complex. Thanks for the lesson. I've heard it stated the hdv takes 7 times more processor power. I'm just curious though, how are all the people using the hd100 editing? In real world time, how much more time consuming is it working with hdv? Just wondering, thanks
Click to expand...

I own an HD100, though I haven't even seen a real HVX200 I'd guess the answer is much to do with the respective NLE/hardware. The best way to edit HDV is to convert to an intermediary codec. Premiere and cineform seem to provide a reasonable workflow as does Edius and HQ. Not sure about Vegas.

Also, keep in mind, JVC's flavor of HDV uses only a 6 frame GOP so the shorter string of frames probably mitigates some of HDV's deficiencies. To illustrate, here's what JVC's GOP looks like:

IMPEACH
IMPEROR
GWB
B4
HEWRECKS
AMERICA

But no matter, DVCpro remains the more robust container.

Hope this helps.
 
When I had my Sony FX1 (2 years ago... sold it) I spent a lot of time fondling it and marvelling at the futuristic design... the way a person admires a beautifully crafted hammer.

I acutally limited my shooting time so I can keep my video heads pristine as possible... perhaps to delay the day I see the dreaded drop-out.

Now that I have the HVX, I spend a lot of time marvelling, not the camera, but at the awesome images the HVX generates.

It is a black and solid hammer that screams: "USE me, dammit! Value me not for my looks but what I will deliver..."

In the end, it's the beautiful footage that only matters.

PS: I think I'm developing an "HVX eye". I beginning to see zebra stripes on highlights even when I'm not using the camera.
 
You must log in or register to reply here.
Back
Top