I really dislike the whole shaky cam thing. Fast, creative camera movements are so much more useful. I completely agree Spiff. I want to *see* the fight in good detail. Maybe that's just the martial artist in me though. The best fight scenes are those in which the talent of the combatants is obvious.
Explain 1080P
- Thread starter braw
- Start date
Yea I am not a huge fan of the fight footage from the bourne movies either, or really any american action film's fight footage (I am too much a fan of the Shaw Brother's movies to enjoy it), though I do think the shaky cam can have it's place. I love the moving and shaky camera style of the new Battlestar Galactica series, and think it really ads to the show.
Bourne shaky-cam
Bourne shaky-cam
I really like Bourne Identity but Paul Greengrass just got too excited on Supremacy. He loves that style! It isn't quite so bad on the small screen but the geography of the fight and chase scenes is terrible!
I'd pick John Frankenheimer or Mctiernan as my favourite action directors. I find the Woo School unengaging and false, although I know it is supposed to be unreal. IMO Ronin still has the 2 best car chases ever. Spielberg did pretty well is Ryan and Scott with Enemy of the State... IMO
Maybe we should start an action cinematography string!??
Bourne shaky-cam
I really like Bourne Identity but Paul Greengrass just got too excited on Supremacy. He loves that style! It isn't quite so bad on the small screen but the geography of the fight and chase scenes is terrible!
I'd pick John Frankenheimer or Mctiernan as my favourite action directors. I find the Woo School unengaging and false, although I know it is supposed to be unreal. IMO Ronin still has the 2 best car chases ever. Spielberg did pretty well is Ryan and Scott with Enemy of the State... IMO
Maybe we should start an action cinematography string!??
PixelMagic
Active member
Whoa, hey, Ryan. What's up? I didn't know you were a member of this board.
PixelMagic
Active member
Yep, this camera is gonna rock.
stokestack
Active member
Hm, I'm perplexed about recompression.
Hm, I'm perplexed about recompression.
Hi Barry.
I wonder about the compatibility of these statements:
"Same with the HVX -- it records 1080/60i only. So the progressive-scan footage will be carried within a 1080i data stream: 1080/24p, 1080/30p, and 1080/60i are all recorded within a 1080/60i data stream. Just like with the DVX, 24p will be recorded using either 3:2 or 2:3:3:2 pulldown, your choice.:
then...
"DV (and its variants, DVCPRO-50 and DVCPRO-HD) all employ two modes of compression, based on how much difference the codec sees between the fields. If there's a lot of temporal offset, the codec will compress as fields. But if it doesn't detect much offset, it'll instead compress the entire frame at once."
There should be zero temporal offset with this camera in 1080 24P mode, since the 1080 frame is only being split into "fields" for data-rate reasons. So how can it be called 1080i? How can the encoder, which operates on 8x8 tiles, do so when every other line is missing from each field?
The only way I can see would be for the codec to process the entire frame and then split the data in half and write it out as two half frames that couldn't stand on their own. Similarly, instead of really interlacing them, it could write the top half of the frame and then the bottom. That would solve the every-other-line problem.
Do you know what the codec does?
Thanks!
Gavin
Hm, I'm perplexed about recompression.
Hi Barry.
I wonder about the compatibility of these statements:
"Same with the HVX -- it records 1080/60i only. So the progressive-scan footage will be carried within a 1080i data stream: 1080/24p, 1080/30p, and 1080/60i are all recorded within a 1080/60i data stream. Just like with the DVX, 24p will be recorded using either 3:2 or 2:3:3:2 pulldown, your choice.:
then...
"DV (and its variants, DVCPRO-50 and DVCPRO-HD) all employ two modes of compression, based on how much difference the codec sees between the fields. If there's a lot of temporal offset, the codec will compress as fields. But if it doesn't detect much offset, it'll instead compress the entire frame at once."
There should be zero temporal offset with this camera in 1080 24P mode, since the 1080 frame is only being split into "fields" for data-rate reasons. So how can it be called 1080i? How can the encoder, which operates on 8x8 tiles, do so when every other line is missing from each field?
The only way I can see would be for the codec to process the entire frame and then split the data in half and write it out as two half frames that couldn't stand on their own. Similarly, instead of really interlacing them, it could write the top half of the frame and then the bottom. That would solve the every-other-line problem.
Do you know what the codec does?
Thanks!
Gavin
Barry_Green
Moderator
For three or four out of every five frames, that's true. But in one or two of the frames there will be significant temporal offset. In 24PA mode, one frame out of five will consist of a field from each of two different frames; in 24PN mode two out of every five frames will be made from fields from different frames.stokestack said:
Essentially, yes, but also for compatibility with the codec.
Because it's being split into, and displayed, as an interlaced data stream. If you plug the camera into an HDTV, the TV will see an interlaced/1080i stream coming at it.
This part I don't understand. What's missing? There's nothing missing at all. The 1080p frame consists of 1080 scan lines. A 1080i "frame" consists of 1080 scan lines. The difference is when those scan lines were acquired -- in 1080p, they're acquired all at once, and in 1080i half of them are acquired in one 60th of a second and the other half are acquired in another 60th of a second, but when all is said and done, each mode still has 1080 lines.
It's like if someone gives you a $10 bill, or they give you two $5 bills. Either way you still get ten bucks.
There isn't an every-other-line problem, it's presented all at once, same as DV. However, the method you're describing sounds a lot like how Sony writes its 24pSf HDCAM data -- they segment each progressive frame into half-frames and write those out as fields, with no pulldown. Sounds like what you're describing.
The HVX and DVX and XL2 and SDX don't do that -- they convert 24 progressive frames into 60 fields, and compress that within a 60i stream.