Question about the Nano Flash and my EX1
 

I will be doing nature/wilderness related shooting with my EX1. One thing I've noticed about my EX1 is that even when I do a very slow pan on a dolly, such as using a deep DOF including foreground grasses/flowers and scenic background, as soon as I move the camera it loses sharpness on the foreground objects. I've done a lot of practice footage with my EX1 but I'm new to videography and have never seen how other codex renders motion like this... is it just "video" in general, or is it the Long GOP codex? If it's Long GOP, will using the Nano Flash completely eliminate this issue? Does it matter if I shoot Long GOP or I-Frame, or is the less compression the main factor so either would be fine? Thank you!

Hi Buck-

It's difficult to determine if this is caused by the camera sensor or the compression. The best test is to view the live video out the HD-SDI port and see if the problem is still present. This live view will eliminate any CODEC issues as the video is taken directly off the CMOS sensors. Ideally, if you can capture this uncompressed stream, you can examine on a frame by frame basis.

If it looks good in uncompressed form, then it will likely look good captured at 100 Mbps using the XDR/nano box, as this high data-rate works well even with lots of motion (in our experience).

Hi Buck, in addition to Mike's comments about the codec, be sure and check your shutter settings. First, is your EX1 shutter set to "off"? If so, this can effectively double the blurring you see in any motion scenario, as the shutter duration is the maximum possible for your frame rate (i.e. 1/24th sec for 24fps, 1/30th sec for 30fps).

For "normal" motion blurring you want 1/48th sec for 24fps; 1/60th sec for 30fps.

Regards,

Jim Arthurs

Buck,

I am wondering the same myself.

I don't have an EX-1, but this is a known issue with the EX-1 compression, and higher bitrates might help.

But and I-frame setting would solve the issue as each frame gets processed like you are shooting a still.

I noticed some of this during the olympic coverage this summer.

I don't know what exact cameras or codec was being used, but when the camera moved, everything got a bit soft until the camera stopped, then everything got sharp again.

This is with slight movements, not fast, beat the shutter speed type of movements.

Panasonic provided more than 100 HD cameras (HPX3000, HPX2000)
Sony had 78 cameras (28 PDW-700s, 42 HDC-1400s, HDC3300 (slo-mo cameras), 4 PMW-EX1s)
Thompson Grass Valley provided 550 cameras (LDK 8000 Mk2, LDK-6000, LDK-8300 super-slomo,


Codec were: DVCProHD, XDCam HD, XDCam EX, JPEG2000 (TGV)

So, if you saw motion breakup, you were seeing it with some of the best hardware money can currently buy right now.

Source: HDVideoPro October 2008, pp. 46-51

Dear Tim,

How were you watching the Olympic coverage?

For example, was it via cable or live over the air HD broadcast?

The cameras could have been perfect, but by the time you saw it, it could have been highly compressed for transmission to your television.

I was watching over DirecTV HD on a Panasonic 1080p Plasma.

The area where I noticed the out of focus on movement was during the gymnastic coverage.

For example, when gymnasts were stationary waiting for there results, a camera often on a steadicam would move over to them.

When the camera was in motion or even if there was a slight adjustment, the frame would go a bit soft.

When the motion stopped, a definite return to clarity happened.

I may be wrong, but I thought Sony equipment was being used in the gymnastic coverage and Panasonic was used in the swimming/diving and anchor desk.

I think I saw Sony looking camera bodies in the gymnastic area.

Anyway, the Panasonic folks say this is a big drawback of the EX-1 family of cameras as the codec needs a lot of bandwidth to not eat away at the scene.

Not rip on any camera, but I did notice this exact effect during the olympic coverage.

The bitrate in the EX1 is fixed, however, the compression rate is variable. How can this be, a curious mind may ask? The long GOP compression uses whatever level of compression it needs to meet the 35mbps rate. When the cam is held still and motion within frame is small, practically the entire 35mbps rate is available for GOP's . When you pan the camera and the frames see "drastic" motion, a lot "I" frames are needed and more of the bandwidth used for GOP compression is used for I frames. The result you see is a slight blurring of the image. not blocks or pixelation, but blur...looks a lot like motion blur.

Caveat: My use of the IGOP nomenclature is metaphoric. It's the best way I can describe how the Sony HQ codec works.

Thank you for the comments/suggestions. I haven't messed with the shutter, not even sure if it's on or off, so I'll have to play around with various scenarios and combinations to see if I can solve the issue. If the Nano Flash will solve it, they've got my moolah right here and now. But admittedly I don't wouldn't know a bit rate if it came up and bit me. I learned photography on my own shooting gazillions of rolls of film, and it's paid me well once I figured it out. With video I guess I better actually crank open some manuals and learn this stuff, it's a lot more techy than stills. I could eventually learn on my own like I did with photography, but by then all this new gear I've bought will be outdated, so I need to expedite my learning curve with studying. But all this video tech stuff is like learning an el foreign-o language-o to me-o.

Buck, if I remember right, the camera ships with the shutter on/off switch in "off". Check the front of the camera, below the lens for the location of the switch. Turn it on, verify in the menu that the shutter is appropriate for your frame rate as outlined in my previous post and compare the results.

Regards,

Jim Arthurs

Gracias, Jim, I will do that for sure!

Okay, I tried various settings (and my shutter is on). I tried 1080/60i/30p/24p and 720/60p/30p at the default shutter speed (such as 60 for 30p, etc.) and I tried 1080/30p at a faster shutter speed (250).

I used a row of books on a bookshelf as my test and slowly horizontally panned slowly across the book spines. The motion breaks up the sharpness noticeably. The best in my unscientific test was 720/60p, but it still wasn't good. Once I stop the camera everything is razor sharp... once I slowly pan it immediately loses sharpness. I viewed this on my 30" Apple Cinema Display... I haven't had time to determine if it's the monitor refresh rate and maybe the motion would stay sharp after going to blu-ray or whatever, but I'm not confident about that.

Are there any plans soon to test "motion" with the Nano Flash? Just basically doing a quick and simple EX1 pan with and without the Nano Flash (or XDR)? Motion on static subjects is my only concern here. I'm happy with the color and sharpness of stills (or even following a moving subject), but the visions I have in my mind of what I want to create on video defintely requires sharp rendering of motion on static foreground scenes. El periodo.

I dont think codec
 

has much to do with what your referring to here, sure you get codec breakup etc, but any pan will smooth things out so I'm not sure why your looking for sharp detail under these circumstances? you can up the shutter speed quite high to improve this.
Or maybe I'm just missing the whole jist of what's being discussed here? Which as Perrone said the bit rate won't help change this fact.

Sorry if I've missed it??

Adam

OK, I'll say this as simply as I can. What you're seeing is n-o-r-m-a-l for 35mbps encoding with the EX1. Maybe going to a higher bitrate(like 100mbps), will help this out. At least, I'm hoping so, but since I don't have my nanoflash yet, I don't know for sure. Since the EX1 is limited to 35mbps in HQ mode, recording on nanoflash via the HD-SDI port is one way to get higher bitrate recording. Shutter speed won't help this, frame rate or format will help only to the extent that certain frame rate and format settings (like 720p24) ease the compression crunch. But then again, panning in 24fps has other issues. What would help the most is to stop panning.

Hey Doc, it only hurts when I laugh.

see this post:http://www.dvinfo.net/conf/convergen...ke-schell.html

I think the main thrust of this thread is wondering if a higher bitrate will diminish or take this effect away.

But the nice thing about the Flash XDR is that it also has an I-Frame option which will definately take this away.

Which is what I will be using for shots with a lot of motion or panning etc...

Tim...yes, thanx for the clarification.

Gracias for your patience in explaining this, ha! Like I said, I'm new to this video stuff so I don't know what "normal" is. Some people are telling me it should remain sharp with s-l-o-w pans, and some are saying it's normal. Some people keep asking for "motion" tests with the XDR/NanoFlash because they imply motion is the weakness of Long GOP. So I don't know what normal is due to lack of experience, or if intra-frame doesn't have this issue, or if Long GOP wouldn't have this issue if the bit rate was much higher, etc. If I-frame will solve the issue, as was just suggested (if I read it correctly) then that would suggest it's a Long GOP compression issue. I guess. Maybe. I think. ha!

I guess I'll just hang out and be patient and wait for some motion testing with the XDR and/or NanoFlash to see if there's a difference. Thank you!

yeah....kinda what I'm doing....

When I first saw this "blurring" on pans with my EX1, I assumed it was because I was panning too fast, as others have suggested. I practiced with excrutiatingly slow pans, in my attempt to minimize the motion blur, to no effect. Those who suggest this is an artifact of panning too fast have no experience with the 35mbps Sony encoder. What I've seen is that anytime the entire frame translates horizontally (not necessarily vertically), this faux motion blur happens. It's really not motion blur, but, compression artifacting due to motion.

Buck and Bill I hope I can answer both of your questions when I receive my XDR Flash in a month. Most of what I shoot is fast motion and hoping this will improve picture quality at 1080p 30p 100Mbps 422.

Hi Adam! What I'm referring to are very slow pans. I watch nature HD flicks a lot and they are magically sharp even with slow panning movements. I'm not seeing this with my EX1... it's razor sharp until the slightest movement. When I shoot 108060i this is minimized, or 720 60p, so I'm wondering why I can't shoot 1080 30p and keep it sharp with slow pans? What's causing it to lose sharpness when I don't see it on HD programs, or nearly as much with 1080 60i. I'm still learning all this stuff. I have no idea why.

Buck after a long conversation with Mike and reading everything I could on the subject in a week I am convinced the faster bit rate will help. How much I do not know. And like you I do not know the details but being an engineer it all makes sense to me. Still this does not mean I fully understand it so only one way to find out.

It will definitely help if it isn't an EX1 image processing issue. I know that I get much sharper rendition of motion when I've recorded to higher bit rate codecs via SDI from the XL-H1. (shutter speed notwithstanding...obviously how sharp a pan looks does depend somewhat on pan speed and shutter speed)

I've just had a sneak peek at a 100mbs XDR-processed version of a ProRes HQ raw SDI clip I submitted to C-D.

The result is stunning. The visual fidelity of the 100mb MPEG2 422 mode is basically the same as ProRes.

There's no doubt it would be a signifigant improvement over XDCAM EX HQ compression. The added color resolution alone is worth the price of admission to me.

Thanks Barlow

Ask and ye shall receive.

Here is a comparision between a frame shot with the EX1, 24fps at 1/2000th second shutter, recorded on both the camera's 35Mb/sec codec and the FlashXDR at 100Mb/sec.

Why the super high shutter speed? I want to remove any trace of motion blur from this test. The two ways you can stress a codec are to have high detail in the picture and fast temporal changes between those images. So I shot a whip pan with razor sharp images that change significantly from frame to frame...

Here is the source sequence for reference..

http://ftp.datausa.com/imageshoppe/o...astshutter.mov

And here is a 200% enlargement of one of those whip frames, with heavy levels applied to pop out the compression blocking...

http://ftp.datausa.com/imageshoppe/o...ighShutter.png

Pretty clear that the FlashXDR's 100Mb/sec provides the more robust picture under "codec buster" situations. Just look at the macroblocking in the tree, the ragged edge of the tree trunk and the chroma blur of the trailer's red lights in the 35Mb/sec image.

And here are are the 35Mb/sec and 100Mb/sec frames as a Photoshop layered file for any who wish to do their own image manipulation tests... of course you'll need Photoshop and understand how to work with a layered file...

http://ftp.datausa.com/imageshoppe/o...ighShutter.zip

Regards,

Jim Arthurs

Hi Jim-
Thanks! I studied the layered Photoshop file in great detail. There is a very clear reduction (essentially eliminination) in the compression artifact level in the 100 Mbps 4:2:2 over the 35 Mbps 4:2:0. Admittedly, this a real torture test of a fast panning shot, but it really shows how well the long-GOP format holds up to high-motion content, especially as you go to the 100 Mbps level.

I am now convinced that there is little or no gain going to the 160 Mbps I-Frame only mode. While we will offer this mode, I think the video quality will be inferior to the 100 Mbps Long-GOP, even in the most demanding scenes. The Sony engineers did a brilliant job with this long-GOP CODEC design, all you really need to do is crank up the bit-rate to get visually lossless video, IMO.

It is very clear that 100 Mbps Long-GOP Quality >> 100 Mbps I-Frame, especially DVCProHD, which subsamples the video to 1280x1080 before compression. Comparison reports from German testers have indicated that I-Frame Only 100 Mbps JPEG2K is below the 35 Mbps HQ mode (in the EX1/EX3) in overall quality. Again, the ability to compress redundancies within a Frame (I-Frame) and from frame to frame (P,B) can produce superior video quality over the simpler I-Frame only CODECs.

This is really no big surprise, but it's great to see real-world evidence.

Hey Mike...

I know you probably get harassed all the time about schedules. Guess I'm no different. Really wondering what the latest projection is for release of nano-flash to early adopters.

Also, since I'm mainly a PC type user, will Avid be able to handle the 100mbps sony codec. It does quite well with RED and DNX185. Is anything in the works for Avid and nano-flash?

TIA

I think the 100mbs Long GOP is actually superior to standard ProRes, and maybe even equal or better than ProRes HQ, given an 8-bit quantized image.

Hi Bill-
No problem asking about the progress. We're in PCB (Printed Circuit Board) layout and cabinet design. In all honesty, progress on nanoFlash is behind schedule, primarily due to debug / development of the Flash XDR. The good news is that 95% of the code will transfer from XDR to nano, so once we can transfer our engineers off XDR, we should be able to finish nanoFlash in very short order.

We should start debug of nanoFlash in November and hopefully have some initial units in December. I'll keep you posted on the progress. I apologize for the delays.

Hi Bill-
I forgot to answer the second part of your question..

We are hopeful that Avid, Edius and Vegas will be support the 100 Mbps data-rate. Based on our experience with FCP, I think this is very likely. We should know in about 2-3 weeks.

Great news, thanx.
I follow the updates on Flash XDR. Good stuff, especially since I know that Flash->nano is a small step. The hardware is another thing, which you filled me in on. That's what I was asking about. Thanx for the update. I am patient, no hurry.

Mike, I was in the market for one of these units because I needed a better long form recording solution than the SxS cards. I really didn't need the 100mbps capability, though it was going to be a great side benefit. Given the SDHC solution, I cannot now justify the purchase.

That said, I think these boxes you guys are building are remarkable. It will open these cameras, to an entirely new audience of users. It should help push the EX1/EX3 out the door in droves next year. Those in the industry looking for a cheaper solution to SDI recording will no doubt be thrilled. Especially once you get he uncompressed recording going. It's overkill at our level, but for the big boys, a necessity.

I wish you guys all the luck in the world with these, and maybe someday, I'll be able to convince the powers that be here to go that route. But it will be some time before I get there!

All the best.

Jim, thanks for the test! I'm more concerned about very slow, very sharp pans, as you provided in the other forum. Since I don't know much about specs, and it appears the 'specs' sound good from everyone talking about them, my eyes still see distracting motion blur.

I'd be willing to drive to Colorado Springs are wherever Convergent Designs is located to "test" a NanoFlash before buying, if they allow it (when it's available). I'd hate to plunk down $3,500 if it doesn't truly solve my issue, especially since that is probably more than the price of a new Scarlet (whatever that camera will end up being, who knows!).

In the meantime I'll trying shooting more 1080/60i and see if that helps my motion issues. Thanks again!

Hey Buck,

The issue you are speaking of is pretty clear in this thread.

It is solely dependent on the recording codec and bitrate.

This is not motion blur and is not effected by framerate.

In short, this is the recording codec being overloaded with information (with camera movement, every pixel changes from one frame to another, therefore no redundant pixels can be compressed) and taking image quality away to be able to function.

When movement stops, the amount of changing information goes way down and the codec does not have to cannibalize its image to keep up and sharpness returns.

The Flash XDR will give the codec more headroom (100Mbps vs 35Mbps) to handle more information, and hopefully allow it to never have to go so far as to lose detail and sharpness.

Hope this helps!

Buck has this thread going on several websites. He just doesn't seem to want to hear the message. If he wants to drive to Colorado Springs, and it makes him feel better, let him....;o)

Ha! I asked the same question on one other site because I don't know who goes where and there seems to be different quantities/qualities of advice on each site. I also don't see the same screen names at both sites so I don't know about the redudancy. I apologize if everyone visits both sites and I'm being redundant. But my redundancy is on dos sites, not 'several'. If there are other sites I don't know about that I can ask this question, please fill me in, I'd be glad to ask there too! :^D

As for "getting it", I'm not seeing a consensus on this issue? Some skilled videographers are telling me that the EX1 codec is fine for motion and what I'm seeing is not the limitations of the EX1, but the limitations of the LCDs I'm watching it on (refresh rates, etc.). They say if I output to Blu-Ray or DVD I won't be seeing these motion artifacts.

Others are saying it is the limiations of the EX1 codec and the NanoFlash will alleviate the problem. I need the problem to vanish, not just be "less", so I don't know if I want to spend $3,500 for something that will just make it 'less' of a problem.

I'm not trying to be difficult! I'm new at this stuff and I am admittedly ignorant on all this tech talk. With the quick-pan test above I actually have no idea what I'm really looking at... I actually see bad blocking in both, one is just a little less than the other, but I wouldn't do such pans anyway so it's not a concern. I'm talking very slow, creeping horizontal pans with a stationary foreground object. I have yet to see anything 'visually' acceptable, on this site or the "other", that would fit my 'vision'. Admittedly maybe what I'm trying to achieve can't be done without a $100,000 three 2/3 CCD camera? I was hoping maybe the NanoFlash will be my "savior", ha! :^D

Hi Buck-
We should have some very definitive comparison shots this afternoon, which compare 35Mbps 4:2:0 to 100 Mbps 4:2:2. The difference is very clear.

Hey Buck,

We are just trying to help!

I am not an EX-1 owner, I really considered it but the rolling shutter and lack of an affordable "Big" camera to go with it steered me to Panasonic.

Plenty of people are more than happy with the EX-1 and that is fine.

But you are wanting the highest level of output.

If you look at broadcast output, the use of compression is often limited to the last part of the signal chain - the path to your house.

The XDR will allow you to operate more towards this method of production.

Your recorded image will be closer to uncompressed, which will get the best out of your $6,000 camera that you can get.

Which is a bargain. $11,000 or $9,500 for full raster, near uncompressed output.

35Mbps 4:2:0 vs 100Mbps 4:2:2 Comparison
 

Here's some great comparison shots courtesy of Jim Arthurs. The video was simultaneously captured on his Sony EX1 using the internal 35 Mbps CODEC in HQ mode and externally via HD-SDI into the Flash XDR. The Flash XDR CODEC was set to full-raster 1920x1080 4:2:2 Long-GOP at 100Mbps.

The images were taken from a horizontal pan with a very high speed shutter setting to eliminate blur.

Zoom in and you will clearly see the differences. The 100 Mbps is virtually 100% free of any compression artifacts.

IMO, the 4:2:2 Long-GOP format holds up extremely well at 100Mbps, even in high-motion scenes and is virtually indistinguishable from uncompressed.

Thank you, Tim, and I do understand and appreciate your help. Bill just seemed to imply I was posting this on 'several' forums and was burying my head in the sand and not listening. I am eager to learn and I am taking all considerations and I'm taking notes and am certainly not "not wanting to hear the message". Of course it's all in fun. We're just talking cameras here anyway.

I'll admit I take things Jan says with a grain of salt, even though she is no doubt knowledgeable, she works for a company that doesn't really want to say good things about the EX1, ha! :^D That would be like asking Obama what he thinks of McCain's campaign. I'm wary of possible agenda-driven advice, no matter how good it may be.

You say it's codec driven, and I have no reason to doubt that. Someone else says it's LCD refresh rates and the codec is fine, and I have no reason to doubt that either. That's just my lack of experience. I'm gonna have to "see" it myself. That's why I would be happy to drive to Colorado Springs... it's a nice drive anyway and I love that place, plus I don't want to spend that kind of moolah and "hope" the results are what I want. Mike said he'll have some results up soon so I'll see if it's the kind of footage and results that I hope will help me here. I have no doubt the higher bit rates will be a big improvement spec wise, but like I said, I judge with my eyes not tech specs. If it's just the nature of "video" and motion, then that's just how it is I guess, but I see stuff that tells my eyes motion can remain sharp with movement. But that may be $100,000 talking too, ha ha!

I honestly do appreciate your help. My day job has been keeping very busy (as well as my kids!) so I haven't had much time in the last week to thoroughly test the 60i option (I've been shooting 30p), but this week I should be able to do more. I'd like one of those Sony F900's... since the government seems to be handing out moolah all over the place, maybe they'll be so kind. I think I'll change my company name to AIG Productions. :^D

Hey Mike, cool, and thanks... you don't happen to have any "motion" comparisons, such as slow pans, do you? Let's say 1080/30p at 35 mpbs, and then at 100 mbps? Stills don't tell me the story my eyes want to hear. Not that I expect you would do so just for me, I'm just wondering if you have any in all the testing you do with this thing.