over/under crank

Steve Phillipps · August 1st, 2009, 01:53 AM

These fast cards are expensive but I think you just have to keep telling yourself that they'll last years and that compared to tape they'll still work out cheap. It's got to be the best idea to put the most bomb-proof cards in the camera when we're working so hard to shoot al this great stuff.
My plan would be to buy some slower cards (Sandisk III) to do 100 mb/s long GOP for 25P stuff and a couple of faster ones (sandisk IV) for 160 or 220 mb/s stuff for slomo.
Steve

Ofer Levy · August 1st, 2009, 08:21 PM

Hi all,

Tried to get the slow motion by shooting 720 60p at 160 Mb/s I-Frame only then trying to conform in Cinema Tools to 25 fps as Steve suggested- no luck.
Used 3 different cards: Transcend 32 GB, Sandisk Extreme IV 4GB and Sony 306x 46 MB/s 8GB.

I get the same message in the conform log:
"Skipped - The movie file is not writable .........MOV
Unable to move the file.........Mov"

The clips playback fine but still fail to be conformed.

Any ideas ayone as to what can be the cause for this failure?

Thanks,

Ofer

Dan Keaton · August 1st, 2009, 10:19 PM

Dear Ofer,

1. The Transcend 133x 32 GB card is not rated for 160 Mbps.

2. Officially the SanDisk Extreme IV 4 GB card has not been tested.

3. Officially the Sony card has not been tested.

I recommend that you test the 140 Mbps I-Frame Only mode with your Transcend 133x 32 GB card.

Ofer Levy · August 1st, 2009, 11:17 PM

Hi Dan, the clips are plying back fine.

Will do the test with the Transcend 32GB at 140Mb/s tomorrow and report to the forum.

Thanks,

Ofer

Steve Phillipps · August 2nd, 2009, 02:24 AM

Ofer, you need to ingest the EX clips as usual, but then they need to be transcoded to an I Frame codec. In FCP go to File, Export, Quicktime Conversion. Then choose Apple Prores HQ, set your destination etc. This will then make a Quicktime at your destination, open this with Cinema Tools and conform.
Do this on a few clips one at a time to begin with to see how they look, but you can also do whole cards by File, Batch Export, and then by Batch Conform in Cinema Tools. The CT bit is quick, but the transcode to ProRes is time-consuming, and worth setting up to run overnight.
Steve

Dan Keaton · August 2nd, 2009, 04:05 AM

Dear Ofer,

When a chip is inserted, we attempt to identify the Brand/Type/Speed of the chip by reading information built in the controller of the chip.

Some original, or early, Transcend cards do not have this information.

As such, we can not always detect that one is using a Transcend 133x 32 GB card.

Thus, we can not lock out the 160 Mbps modes on these cards, as we do for other cards.

We do not recmmend recording at 160 Mbps onto a Trancend 133x 32 GB card. We do expect this to fail, but it is not an instant failure, it is a failure after a period of time; after our write cache fills up, since the card can not accept our writes as fast as necessary for 160 MBps.

Ofer Levy · August 3rd, 2009, 12:53 AM

Hi all,

Tried it again today - used the Transcend 32GB, shot 720 60p, I-Frame only at 140Mb/s.

Got the most stunning slow motion!! Way superior to the slow motion I normally get with the EX3.

I actually realized the only thing I had to do differently is to open the clip in FCP first, save it and then open the saved file in Cinema Tools in order to conform to 25fps.

Will get my Sandisk Extreme IV 16 GB in a few days and will try it with 220 Mb/s.

Thank you Dan for the fantastic customer support and a HUGE thank you to all the team in Convergent -Design for this absolutely amazing device. The footage I get using the Flash XDR is definitely in a different league to what the Sony PMW EX3 can natively produce.

Cheers,
Ofer Levy Photography

Tim Polster · August 3rd, 2009, 07:46 AM

Dan,

Given the success of I-frame for slow motion, would it be possible to add a 50mbps I-frame setting?

Sometimes longer record times are needed and for many uses 4:2:2 50mbps I-frame or long GOP is fine quality.

BTW, does this I-frame/long GOP difference apply to chromakey?

Dan Keaton · August 3rd, 2009, 07:58 AM

Dear Tim,

We could run a test, but I am initially concerned about the quality of 50 Mbps I-Frame Only.

Yes, for chroma key, we recommend using 100 Mbps Long-GOP or the high bit-rate I-Frame Only modes.

Tim Polster · August 3rd, 2009, 08:40 AM

Hey Dan,

Hope you are well.

Does the Long GOP or I-frame have in impact on chromakey like it does on slow motion outside of bitrates?

Since having every frame helps slow motion I wondered if keying software might benefit from having every frame as well.

About the quality of 4:2:2 50mbps I-frame, I am guessing but I would imaging that it still would look better than 4:2:0 35mbps long GOP.

The usefulness might outweigh the image quality difference.

Dan Keaton · August 3rd, 2009, 08:51 AM

Quote:

Originally Posted by Tim Polster

Hey Dan,

Hope you are well.

Does the Long GOP or I-frame have in impact on chromakey like it does on slow motion outside of bitrates?

Since having every frame helps slow motion I wondered if keying software might benefit from having every frame as well.

About the quality of 4:2:2 50mbps I-frame, I am guessing but I would imaging that it still would look better than 4:2:0 35mbps long GOP.

The usefulness might outweigh the image quality difference.

Dear Tim,

One common misconception is that Long-GOP does not produce every single frame for the Non-Linear Editor. I had this same discussion with an Apple expert at ROSCOR 2009 event in Chicago. He did not believe that Long-GOP produced every single frame.

He thought that if we were filming a hockey match, we could not see the hockey puck, unless it was in an I-Frame in the Long GOP structure.

To prove that it did, we created a test file for him with very fast lateral motion in the frame.

He was amazed at the test results as he did not think Long-GOP could capture each frame of video, but it does.

(His comment was "That is the best video I have ever seen!", then he realized that he had to be politically correct so he said with a big smile "That is among the best video I have ever seen!". It was a great moment.)

Long GOP, when decoded by the Non-Linear Editor, provies every single frame to the NLE.
And since we are full raster and high-bit rate, we provide very high quality images to the NLE.

So, yes, Long-GOP, due to its efficiency is very well suited for chroma keying.

We will need to run tests to see if the quality of 50 Mbps 4:2:2 is acceptable for some uses. Remember that Long-GOP has at least a 2 to 1 or 2.5 to 1 efficiency advantage.

Also, please remember that we are neutral in that we offer both I-Frame Only and Long-GOP, via a menu selection, in our Flash XDR and nanoFlash. It is not as if we are attempting to sell one over the other, we support both.

But, we do know the advantages of each, and that 100 Mbps Long GOP is far better than 100 Mbps I-Frame Only. 100 Mbps Long GOP and 220 Mbps I-Frame Only are probably close.

Tim Polster · August 3rd, 2009, 10:07 AM

Thanks for your reply Dan.

I mis-spoke (typed) as I was just trying to separate I-frame from long GOP.

I realize every frame is there, just some have more info than the others. Should have put every "full" frame...

What would cause the slow motion to be better using the I-frame than the long GOP if all of the frames are represented in both?

Is it more how the NLE handles the GOP rather than the footage itself?

Dan Keaton · August 3rd, 2009, 10:14 AM

Dear Tim,

We are having an internal debate in our company concerning the use of Long-GOP in over and under-cranking.

Until we try Long-GOP with over and under-cranking we will not know if it will work or not, so we are not promising it.

Until then we are expecting to support I-Frame Only for over and under-cranking.

Mike Sertic · August 3rd, 2009, 10:20 AM

Dan,

I plan on frequently using the nano to record at 720p60 and then conform to 720p24. This is in an application where there is a lot of non-linearly moving water with changing surface textures in the frame, which is usually a tough situation for long-GOP compression like HDV.

With that in mind, would you say that long-GOP is potentially more efficient per bitrate/frame at higher frame rates than low, even when there is a lot of motion involved? So that the potential advantage of long-GOP over I-frame only is even greater at higher frame rates?

My understanding is that long-GOP encodes the difference between I-frames and the intervening frames and then compresses that difference, so if there were more frames per unit time there would be smaller differences between frames, so less information relating to frame/frame difference would need to be compressed into the the overall data stream, and more information relating to single frame quality could be captured. If that is a bit confusing, another way of stating my question would be whether the same action recorded at 720p60 at 100mbps would have significantly higher image quality as judged by looking at each individual frame than 720p30 at 50mbps.

If that is the case, I suppose that might be another reason to consider using long-GOP to capture footage to be used in slow motion, unless there is some difficulty in conforming the long-GOP footage in NLEs to a slower frame rate that I am not aware of.

In practice, I expect to test both long-GOP and I-frame only at the highest bit rates my cards can support and see which gives the best results, but it is interesting to speculate. The other consideration with long-GOP at lower bitrates like HDV is that you see a substantial degradation in overall detail / increase in artifacting as motion increases in the same clip (this is especially apparent when viewed in slow motion), and the contrast in quality between the parts with more and less motion can be quite off-putting. It will be interesting to see how the long-GOP fares in this regard as the bitrate goes up.

Dan Keaton · August 3rd, 2009, 10:47 AM

Dear Mike,

Please forgive me, but I preparing for a trip.

I will attempt to answer all of your questions as soon as possible.

But for now....

Our experience with extensive testing has shown that 100 Mbps Long GOP (and Full Raster) is very difficult to break.

Extremely high level of detail does not break it.

High levels of motion in the image does not break it.

High levels of camera motion does not break it.

Mr. Richard Wolnowski tested the Flash XDR with his Thomson Viper extensively.

To test the above, he shot the Houston Marathon, he was in the back of a moving pickup truck, with his Viper, with 25,000 runners coming at him. In other words, lots of detail, lots of motion. This is an extreme case.

In watching the video, it was visually perfect. In analyzing the video frame by frame he found two artifacts (as I remember).

We now offer 140 Mbps Long GOP and 160 Mbps Long GOP in addition to our recommended 100 Mbps Long GOP. This was a torture test. In every other torture test our 100 Mbps Long GOP has worked well, just as it did in this torture test.

Our recommendation to use 100 Mbps Long GOP is based a real-world field testing.

We recommend that everyone test the options available to them, with their choice of CompactFlash card speed, and make the decision that best fits their needs.

The beauty of this is you can choose whats best after you have the unit. You do not have to decide before you buy, as one would have to do with a camera. And, you can switch modes/formats to fit your needs on a shot by shot basis.

August 1st, 2009, 10:19 PM	#18
Dan Keaton Inner Circle Join Date: Dec 2002 Location: Augusta Georgia Posts: 5,421	Dear Ofer, 1. The Transcend 133x 32 GB card is not rated for 160 Mbps. 2. Officially the SanDisk Extreme IV 4 GB card has not been tested. 3. Officially the Sony card has not been tested. I recommend that you test the 140 Mbps I-Frame Only mode with your Transcend 133x 32 GB card. __________________ Dan Keaton Augusta Georgia

August 2nd, 2009, 04:05 AM	#21
Dan Keaton Inner Circle Join Date: Dec 2002 Location: Augusta Georgia Posts: 5,421	Dear Ofer, When a chip is inserted, we attempt to identify the Brand/Type/Speed of the chip by reading information built in the controller of the chip. Some original, or early, Transcend cards do not have this information. As such, we can not always detect that one is using a Transcend 133x 32 GB card. Thus, we can not lock out the 160 Mbps modes on these cards, as we do for other cards. We do not recmmend recording at 160 Mbps onto a Trancend 133x 32 GB card. We do expect this to fail, but it is not an instant failure, it is a failure after a period of time; after our write cache fills up, since the card can not accept our writes as fast as necessary for 160 MBps. __________________ Dan Keaton Augusta Georgia

August 3rd, 2009, 12:53 AM	#22
Ofer Levy Major Player Join Date: May 2005 Location: Sydney-Australia Posts: 443	Good news Hi all, Tried it again today - used the Transcend 32GB, shot 720 60p, I-Frame only at 140Mb/s. Got the most stunning slow motion!! Way superior to the slow motion I normally get with the EX3. I actually realized the only thing I had to do differently is to open the clip in FCP first, save it and then open the saved file in Cinema Tools in order to conform to 25fps. Will get my Sandisk Extreme IV 16 GB in a few days and will try it with 220 Mb/s. Thank you Dan for the fantastic customer support and a HUGE thank you to all the team in Convergent -Design for this absolutely amazing device. The footage I get using the Flash XDR is definitely in a different league to what the Sony PMW EX3 can natively produce. Cheers, Ofer Levy Photography

August 3rd, 2009, 07:58 AM	#24
Dan Keaton Inner Circle Join Date: Dec 2002 Location: Augusta Georgia Posts: 5,421	Dear Tim, We could run a test, but I am initially concerned about the quality of 50 Mbps I-Frame Only. Yes, for chroma key, we recommend using 100 Mbps Long-GOP or the high bit-rate I-Frame Only modes. __________________ Dan Keaton Augusta Georgia

August 3rd, 2009, 10:14 AM	#28
Dan Keaton Inner Circle Join Date: Dec 2002 Location: Augusta Georgia Posts: 5,421	Dear Tim, We are having an internal debate in our company concerning the use of Long-GOP in over and under-cranking. Until we try Long-GOP with over and under-cranking we will not know if it will work or not, so we are not promising it. Until then we are expecting to support I-Frame Only for over and under-cranking. __________________ Dan Keaton Augusta Georgia

August 1st, 2009, 01:53 AM	#16
Steve Phillipps Inner Circle Join Date: Apr 2006 Location: Wales Posts: 2,130	These fast cards are expensive but I think you just have to keep telling yourself that they'll last years and that compared to tape they'll still work out cheap. It's got to be the best idea to put the most bomb-proof cards in the camera when we're working so hard to shoot al this great stuff. My plan would be to buy some slower cards (Sandisk III) to do 100 mb/s long GOP for 25P stuff and a couple of faster ones (sandisk IV) for 160 or 220 mb/s stuff for slomo. Steve

August 1st, 2009, 08:21 PM	#17
Ofer Levy Major Player Join Date: May 2005 Location: Sydney-Australia Posts: 443	Hi all, Tried to get the slow motion by shooting 720 60p at 160 Mb/s I-Frame only then trying to conform in Cinema Tools to 25 fps as Steve suggested- no luck. Used 3 different cards: Transcend 32 GB, Sandisk Extreme IV 4GB and Sony 306x 46 MB/s 8GB. I get the same message in the conform log: "Skipped - The movie file is not writable .........MOV Unable to move the file.........Mov" The clips playback fine but still fail to be conformed. Any ideas ayone as to what can be the cause for this failure? Thanks, Ofer

August 1st, 2009, 11:17 PM	#19
Ofer Levy Major Player Join Date: May 2005 Location: Sydney-Australia Posts: 443	Hi Dan, the clips are plying back fine. Will do the test with the Transcend 32GB at 140Mb/s tomorrow and report to the forum. Thanks, Ofer

August 2nd, 2009, 02:24 AM	#20
Steve Phillipps Inner Circle Join Date: Apr 2006 Location: Wales Posts: 2,130	Ofer, you need to ingest the EX clips as usual, but then they need to be transcoded to an I Frame codec. In FCP go to File, Export, Quicktime Conversion. Then choose Apple Prores HQ, set your destination etc. This will then make a Quicktime at your destination, open this with Cinema Tools and conform. Do this on a few clips one at a time to begin with to see how they look, but you can also do whole cards by File, Batch Export, and then by Batch Conform in Cinema Tools. The CT bit is quick, but the transcode to ProRes is time-consuming, and worth setting up to run overnight. Steve

August 3rd, 2009, 07:46 AM	#23
Tim Polster Inner Circle Join Date: Dec 2004 Location: Arlington, TX Posts: 2,231	Dan, Given the success of I-frame for slow motion, would it be possible to add a 50mbps I-frame setting? Sometimes longer record times are needed and for many uses 4:2:2 50mbps I-frame or long GOP is fine quality. BTW, does this I-frame/long GOP difference apply to chromakey?

August 3rd, 2009, 08:40 AM	#25
Tim Polster Inner Circle Join Date: Dec 2004 Location: Arlington, TX Posts: 2,231	Hey Dan, Hope you are well. Does the Long GOP or I-frame have in impact on chromakey like it does on slow motion outside of bitrates? Since having every frame helps slow motion I wondered if keying software might benefit from having every frame as well. About the quality of 4:2:2 50mbps I-frame, I am guessing but I would imaging that it still would look better than 4:2:0 35mbps long GOP. The usefulness might outweigh the image quality difference.

August 3rd, 2009, 10:07 AM	#27
Tim Polster Inner Circle Join Date: Dec 2004 Location: Arlington, TX Posts: 2,231	Thanks for your reply Dan. I mis-spoke (typed) as I was just trying to separate I-frame from long GOP. I realize every frame is there, just some have more info than the others. Should have put every "full" frame... What would cause the slow motion to be better using the I-frame than the long GOP if all of the frames are represented in both? Is it more how the NLE handles the GOP rather than the footage itself?

August 3rd, 2009, 10:20 AM	#29
Mike Sertic Regular Crew Join Date: Apr 2009 Location: San Diego, CA Posts: 43	Dan, I plan on frequently using the nano to record at 720p60 and then conform to 720p24. This is in an application where there is a lot of non-linearly moving water with changing surface textures in the frame, which is usually a tough situation for long-GOP compression like HDV. With that in mind, would you say that long-GOP is potentially more efficient per bitrate/frame at higher frame rates than low, even when there is a lot of motion involved? So that the potential advantage of long-GOP over I-frame only is even greater at higher frame rates? My understanding is that long-GOP encodes the difference between I-frames and the intervening frames and then compresses that difference, so if there were more frames per unit time there would be smaller differences between frames, so less information relating to frame/frame difference would need to be compressed into the the overall data stream, and more information relating to single frame quality could be captured. If that is a bit confusing, another way of stating my question would be whether the same action recorded at 720p60 at 100mbps would have significantly higher image quality as judged by looking at each individual frame than 720p30 at 50mbps. If that is the case, I suppose that might be another reason to consider using long-GOP to capture footage to be used in slow motion, unless there is some difficulty in conforming the long-GOP footage in NLEs to a slower frame rate that I am not aware of. In practice, I expect to test both long-GOP and I-frame only at the highest bit rates my cards can support and see which gives the best results, but it is interesting to speculate. The other consideration with long-GOP at lower bitrates like HDV is that you see a substantial degradation in overall detail / increase in artifacting as motion increases in the same clip (this is especially apparent when viewed in slow motion), and the contrast in quality between the parts with more and less motion can be quite off-putting. It will be interesting to see how the long-GOP fares in this regard as the bitrate goes up.

August 3rd, 2009, 10:47 AM	#30
Dan Keaton Inner Circle Join Date: Dec 2002 Location: Augusta Georgia Posts: 5,421	Dear Mike, Please forgive me, but I preparing for a trip. I will attempt to answer all of your questions as soon as possible. But for now.... Our experience with extensive testing has shown that 100 Mbps Long GOP (and Full Raster) is very difficult to break. Extremely high level of detail does not break it. High levels of motion in the image does not break it. High levels of camera motion does not break it. Mr. Richard Wolnowski tested the Flash XDR with his Thomson Viper extensively. To test the above, he shot the Houston Marathon, he was in the back of a moving pickup truck, with his Viper, with 25,000 runners coming at him. In other words, lots of detail, lots of motion. This is an extreme case. In watching the video, it was visually perfect. In analyzing the video frame by frame he found two artifacts (as I remember). We now offer 140 Mbps Long GOP and 160 Mbps Long GOP in addition to our recommended 100 Mbps Long GOP. This was a torture test. In every other torture test our 100 Mbps Long GOP has worked well, just as it did in this torture test. Our recommendation to use 100 Mbps Long GOP is based a real-world field testing. We recommend that everyone test the options available to them, with their choice of CompactFlash card speed, and make the decision that best fits their needs. The beauty of this is you can choose whats best after you have the unit. You do not have to decide before you buy, as one would have to do with a camera. And, you can switch modes/formats to fit your needs on a shot by shot basis. __________________ Dan Keaton Augusta Georgia