played around with the EX1 today

D

DanHD

Well-known member
I'm a die hard panny guy, but I rented the EX1 for a quick corporate shoot today, and man, I love this cam. I still prefer the look of my hpx170 but wow, i'll miss the resolution the EX1 provides. And with enough tweaking, I was able to achieve a panny look on the EX1. Overall, I had a great experience with it. I see why you guys love it so much!

Just one question though, I'm editing the EX footage now, and it took a good 10 minutes to conform in Adobe Premeire, so I'm guessing it was converting the long GOP into an I-Frame Codec? Is that pretty standard for NLE, to convert to I FRAME?

I mean, can you natively edit LongGOP, seeing that there is so much information missing?

when you transcode, how does the computer get the missing info back to make 15 complete frames?


Thanks guys!
dan
 
yes you're transcoding.

if you cut in edius, fcp or vegs it's usually much faster. Avid transcodes as well so it's slow on the ingest/conversion.

I haven't noticed any long GOP lag and I cut multicam stuff mostly, so I should

It takes the key frame, then replicates it behind the changed portions of each successive frame until the group is one and the next keyframe is reached, then rinse and repeat.

So actually if you think about it, instead of reading 15 entire frames, it's only reading one frull frame in 15, holding that in a buffer, and spitting it out with the changed portions of the 14 intervening frames. With a good enough decoder, that shouldn't take any time at all.

And don't forget that it's a VBR, not a CBR codec. If all pixels in a frame change from one to the next, then the entire frame is recaptured. It's only when there's no movement pixel to pixel that that part of the frame goes to buffer. Somethng like that. Variable (smart) compression anyway, not constant, like HDV.
 
wow.


much more complicated than i originally thought!


So only the non changing info is placed into the buffer during the 15 frame sequence?



how do you play back HDV or XDCAM EX on, say, a television? Since they aren't real frames, how does a TV with a fized refresh rate play them? Does it just display the first I frame than all the changed information from frame to frame after it?
 
Dan - no, that's only the acquisition format, and possibly an edit format, if you choose.
When you master for broadcast, you recode the final result onto a non-GOP format for output.

I usually make an uncompressed Quicktime file that I then take to an output house for D5, HDCam SR or similar tape outs, or just give the Quicktime file to the client directly.
Then they take it to whatever HD mastering format they want to.

I've mentioned elsewhere that there's around 6-9 other stages of compression and decompression between when they receive my broadcast master and the images hit the final viewer's TV set.

So it's been my personal preference to convert the original imagery as little as possible between when I get it (in post) and when I deliver it out. Probably just a liability/legal thing I have since I'm a boutique house, but I'd rather not get into the codec spec wars if I can possibly help it. Therefore I tend to cut each show as close to its original camera format as I can get it. Which to me means Final Cut Pro, since it speaks Quicktime natively.

Sorry, Brian, not too familiar with Premiere, though I do have a copy on my machine somewhere and could look if you'd like. Just not too fond of the interface, is all.
 
Last edited:
Both Premiere CS3 and CS4 should bring in the EX1 footage without changing the original format. Premiere will analysis the footage to create the waveforms and record information about the clip in the background while you work so that it can edit the video efficiently. Premiere also does this with HDV and other MPEG clips.

Glad to hear you like the EX1. I'm still going back and forth over selling my HVX or keeping it. Hope your edit turns out well.
 
Thanks for all your help
im still unclear on one thing though

if longGOP is the aquisition format, and you're creating one giant 15 picture 'frame', then how do you play it back out of the camcorder, say through the component or SDI out?


does the camera have to, well, make I frames out of everything? or does the I frame just get laid down and then the changed pixels get laid on top of it?
 
Dan,

The camera encodes to MPEG-2 (that's the acquisition format, not LongGOP) and uses the whole I-frame, b frame, p frame process to compress the video data. Then on the DECODE end is where the camera (or whatever output equipment) takes the data and shows you everything. It puts that data back together into something you can see/hear right before your eyes/ears. There is no 15 picture frame it's a "Long G.roup O.f P.ictures."

It's a little too complicated to explain here but basically the I frame is a reference, uncompressed frame of information and the rest of the frames within a 15 frame duration are referencing back to that first frame. If there is a complete change of data (say you pan the camera really hard to a different spot) then a new frame may start. At least in theory or with some MPEG-2 encoders.

If I-frames were everything it'd be an intraframe format such as MiniDV, HDCam or DVCPro HD. The same idea/technology of compressing video to fit onto a DVD or Bluray Disc is the same idea that goes into HDV or XDCam EX.
 
Liquid 7 has edited XDCAM EX native since before the camera was even released...

You just have to export to MXF which is not a transcode but a change of the file wrapper.
 
DanHD said:
does the camera have to, well, make I frames out of everything?
Click to expand...
Yes, the camera has to export a full I-frame.

or does the I frame just get laid down and then the changed pixels get laid on top of it
Click to expand...
What you're describing is how the recording system works. In the most oversimplified description possible, the codec records the I-frame, and then for every subsequent frame it lays down the changed pixels on top of the prior frame.

So in recording, it only records those changed pixels, not the full frame.

But for editing or for display, yes the entire frame needs to be reconstituted into an I-frame before it can be seen or used.
 
wow thanks guys, Trying to get my head around this,


so on the decode end, say when a DVD is displayed on a TV, you're seeing all I frames?
 
I frame is generally a term used in definition of a particular type of frame in a group of picure (GOP) structure for an interframe codec. It generally happens to be the first frame of a GOP which serves as a reference point for the rest of the GOP.

When talking about post decompression (and intraframe codecs), all frames are independant so there is no need to call them I frames, we generally just call them frames (or fields).
 
so when HDV and XDCAM are displayed on TVs RIGHT FROM THE CAM, through SDI or whatever, does the cam decompress and build each individual frames to be played?
 
Thank you barry. I apologize for these ignorant questions I just really didn't grasp that concept at first.



Whew. Intraframe is so much simpler lol =p

In all seriousness though. This is a great camera with impressive qualities. It would be the perfect complement to my 170....oh wait. I'm poor.
 
and also, is there literally and technially no loss of quality when transcoding from longgop to intraframe for editing?
 
no, there can only be loss in transcoding. Intraframe codecs (other than uncompressed) also use compression so information is thrown away again. Some of it will be the already lost and regenerated info, some will be of the origonal 'real' information the origonal codec held on too.

Converting from one codec to another codec that is lossy always introduces more loss of quality. However, in many of these intermediate codecs the loss is small enough not to be off concern to most and the subsequent potential gain in minimising future regenerational losses is of more gain.
 
You must log in or register to reply here.
Back
Top