I doubt that LCD refresh rates have anything to do with sharpness/detail and camera movement.

You may question Jan's motive, but she knows her tech and she speaks the truth.

She is on the inside, we are on the outside.

BTW, I have seen quite a few posts where you were defending the EX-1 choice really hard to people like Jan, now you are looking for more from the camera.

What she was saying is the very reason why you are asking about the Flash XDR, the codec is bitrate starved and with camera movement the image detail is compromised.

In the end, we all have to see how much the XDR long GOP will change the equation, but as I posted before, it also will have an I-frame option that will handle motion without question.

All in good spirit and to pass along information...

Thanks for the shots Jim & Mike.

There is quite a bit of difference between the two sides.

Almost to the point that it looks like two different cameras from a detail and color depth point of view.

One can see the 35Mbps 4:2:0 filling in areas with grey while the 100Mbps 4:2:2 side has solid color information.

Tim, don't get me wrong, I'm VERY happy with my EX1, even with motion following a subject. Even while running at high speeds chasing my 2 year old in the park with no stabilizer, the footage is amazing. I'm blown away. It's *only* in a single specific regard, very slow horizontal pans on stationary subjects is where I have my conerns. I have never seen any rolling shutter effects in 100+ hours of shooting on the run, except up close flash, which I won't be shooting anyway (my boy just happened to get a hold of my wife's point-n-shoot and was snapping away right in my EX1 lens, ha!).

Again, I don't doubt Jan's knowledge, just her obvious bias. I mean, she has to be biased, I don't blame her. It's her job. She has come down very hard on the EX1, but again, she's gotta. If I owned a McDonald's and someone asked me about a Whopper, I'd do my best to portray my Big Mac as better, no doubt. Barry and Jan seemed to be implying that any movement will cause wobble effects (Barry even wrote an article saying any CMOS movement induces wobble, not in my experience at all). As long as there's something to follow, the motion is fine with me.

I was told by another experienced videographer that just as the motion of this text is blurry when you grab the scroll bar with the mouse and slowly scroll the text, that's purely LCD refresh rates causing the text blur. That's how it is when viewing EX1 footage on the same monitor. I have no idea. It sounds rational though, ha!

If I go the NanoFlash route, maybe shooting I-frame is the way to go? Would that pretty much be an end-all for this concern? Maybe at a cost of more disk space? If so, I'll bring stacks of harddrives with me in the field if need be!

Thanks again!

Dear Buck,

Good Morning.

The Flash XDR offers "I-Frame only" recording at up to 160 Megabit per second maximum bit rate (Mbps) recording. This is a great option for those who need to place their priority on speed of editing and cannot, or just do not want to use the "Long Group of Pictures" ("Long GOP") option.

However, possibly other than the full uncompressed option, which is coming later, the Long GOP option at 100 Megabit per second maximum bit rate (Mbps) recording will give you the best quality.

The 100 Mb Long GOP option appears to be visually very close to full uncompressed according to all of the tests that we have run as well as the detailed tests that others have run. These tests have been run with still images, moving images, and fast pans.

At this time, we have not found a torture test (that I know of) that shows a breakdown of the 100 Mb Long GOP codec. There will be many more tests performed. Many of these tests were performed with the Sony EX1 and the Flash XDR.

This is good news for us as well as everyone who wants very high quality images without the burden of working with very large uncompressed files.

While this is obviously a problem also related to the limitation of the codec, loss of sharpness in camera movement is normal even in film originated material. You are taking a solid picture of something passing by the lens. It won't look like the object is standing still.
But naturally, the codec makes it worse.

I think there is more going on here than just the codec. Barry Green's and Mikko followup with waveform testing show it clearly. Part of the issue is certainly the bandwidth limited codec. Of that there is no doubt.

However part of it is down to the rolling shutter. Skew is causing malformations in the signal that are painfully obvious to anyone watching critically.

Does this make the EX1 a bad camera? No. It just means that you aren't going to confuse it for an F900 or even a Varicam when you start moving the camera. It's just as simple as that.

The Nanoflash and XDR will remove one the big impediments to making this camera behave like a $30k camera. But nothing is going to make it behave like a $100k camera. We are just not going to get there. Not with an EX1, not with an HVX or HPX, or any other sub- $10k camera. And if Scarlet comes out with a rolling shutter, we won't get there with it either, although with the problem well understood, Jannard has a better chance of actually bothering to try than most others.

Buck, you're my boy. But if you want the images you see on DiscoveryHD, and NatGeo, sell the ex1 and get a Varicam or F900. Problem solved. You can even connect the nanoflash to them and bypass the expensive tapes. I've seen used Varicams for less than $20k with a lens.

Hi Buck-
Just for clarification, here's the clip from which the 35 vs 100 Mbps comparison images were extracted. It is a fast whip pan in which ever frame is different from the previous one. This is a real torture test, especially for a Long-GOP CODEC.

Take a look at the comparison images again. The 100Mbps side is virtually 100% artifact free. I think this is the definitive test for high-motion and Long-GOP recording. If the video looks great under these conditions, I am sure it will look spectacular under a very slow horizontal pan.

We need to rethink the need for I-Frame only recording. At the 100 Mbps level, all the motion related Long-GOP compression issues are basically eliminated. 100Mbps Long-GOP 4:2:2 Full-Raster MPEG2 Quality >> 100Mbps I-Frame Only 4:2:2 Quality due to the more sophisticated compression algorithm.

Hey, I'm just askin' questions! :^D Thanks for the responses. My EX1 looks like the "big boys" in my untrained eyes when still, I'm just seeing if there's a solution to make it look that way on slow pans too. If there's not, there's not. I'm one to seek all alternatives though in my quest! :^D

Mike, I'm sure the whip pan stills are valuable info, but the only way I'll be able to tell if the NanoFlash is for me is to see footage shot how I'll shoot it. The whip pans don't translate for me. Example, if I want a 4-wheel drive jeep for rough off-roading and the sales guy keeps telling me how fast the jeep can go on the highway, it doesn't matter. To the 'techy' guys they can look at a freeze-frame whip pan and say, "wow!". To me I see a really bad still image that I'd throw away if I took it, but that's just because I go by "feel" and how things look. I'm the creative guy, not the tech guy. It's worked very well for me in photography and admittedly I may have to actually learn tech stuff with video, ha ha! But I still have to "see" results. Which is why I have no problem going to Colorado Springs and actually shooting applicable footage to see if it'll get me the results I'm looking for (if you have a business open to the public, that is, I'm not sure what your setup is). I'm really interested in the NanoFlash for sure. I hope the fact that I have to see something isn't too frustrating for some of you guys, I'm getting a sense of "slap this guy in the head" from some of you, but in a nice way, ha ha!

Hi Buck-
You are more than welcome to come to Colo Spgs and check out a Flash XDR for your tests. There will be no difference in the video from Flash XDR to nanoFlash as they both use exactly the same CODEC and CF card media.

We have a loaner unit ready with battery and CF media.

Mike, thank you and as soon as you get production up and going I'll make a trip out there for some testing, with check in hand! I'm sure I'll be impressed.

Perrone, you would know more than me, but does "skew" or any rolling shutter come into play with the EX1 on a very slow, dramatic, creeping horizontal pan shot from a dolly? Or are you talking more generally?

Based on what I've seen, it appears to come into play any time you're moving the camera, just to varying degrees.

Yikes, if that's the case then the drop in resolution is not from codecs but straight from the CMOS chips themselves and NanoFlash would not solve the el problemo?

since the rolling shutter direction vector is "vertical", skew occurs only with orthogonal pans....for the more artistic minded, this means horizontal pans, only, unless, of course, you're in the habit of shooting the cam at a 90 degree angle, in which case it will be skew in the vertical direction, only.

Now Buck, I know you're trying to understand, but, you're bordering on being ignorant. Skew is not not not even closely related to compression artifacting. Two completely different phenomena. Even RED has the rolling shutter problem, and you can get this feature for over $20K.

With a shutter scan rate of 60Hz, it's not too difficult to see some skew in a frame pan. On the other hand, if you're shooting wildlife with a large telephoto lens, the field of regard is so small that image skew is negligible. Assuming, of course, you have a stable tripod for mounting your videocam. I could go thru some numbers to show you this is the case, but, I know you accept only empirical evidence, so I won't. Just try it.

I'm not ignorant about skew being "horizontal", that's a no brainer. Not even sure why you mentioned it? I think I saw that question on "Are You Smarter Than A 5th Grader" and everyone got it right. :Þ

Bill, I also know skew is not codec related. It's CMOS chips. I'm really lost on why you're saying this because nowhere was it implied otherwise? I know codecs and chips are separate thingies with separate issues. I know RED uses CMOS too. I may be ugly, but I'm stupid. Ha!

Which goes back to my question posed to Perrone... if the drop in resolution I'm seeing on 'horizontal' movement is because of 'CMOS skew' (not codec), then bypassing the native EX1 codec with NanoFlash wouldn't be my huckleberry to solve the resolution drop issue, because it's coming from the source, the CMOS chip.

See the confusion, Bill? Some say, "the EX1 codec is fine, it's what you're viewing it on that creates the motion resolution drop". Others say, "it's the 35 mbps Long GOP used in the EX1, bypass this with the NanoFlash and you should be fine". Others say, "this is just motion and video at work, period." And here Perrone is saying (if I'm reading him correctly, hence my question to him), "this resolution drop is caused by skewing directly from the CMOS chip". Even on very s-l-o-w pans. If so, then no codec will help, right?

Which option is the case so I can stop being ignorant? :Þ

Buck,

It's as I said before. There are multiple things going on with what you're seeing. And because you can't articulate it precisely, we have to look at the entire image of the EX1 versus the cameras you are comparing it to.. namely the Varicam and the F900.

Those are CCD cameras, so the CMOS artifacting and issues are going to be "different" between your camera and those cameras.

You are also shooting long GOP, and that is causing some of what you see. And to bring that in line with the cameras you are comparing to, the Nanoflash or a similar SDI recorder is necessary.

They are using better glass than you are, and that is going to make some small difference as well. Though not nearly as much as the other two.

They are more than likely NOT shooting 24p. And that is certainly responsible for some of what you've seen. Maybe more than anything else. Change to 60i or 30p and you should see a difference.

So there is no silver bullet. Will the nanoflash remove some of the problem. Most certainly. Will it turn your $6k camera into an F900 clone? No.

So to me, you're going to be in one of three places.

1. Don't buy anything else, change to 60i, shoot and be happy.

2. Buy the Nanoflash, and remove the CODEC related issues and live with the rolling shutter ones.

3. Sell the whole lot, and buy an F900.

You pay your money, and you take your choice.

Hi Perrone-
Naturally, we would make the case that the nanoFlash provides the best "bang for the buck", as the EX1 + nanoFlash cost <<< F900 cost, but not substantially less in overall video quality. Or for about 10% of the cost, you'll arguable get 80-90% of the video quality. Not a bad tradeoff.

Perrone, gracias! Just a quick few things and I'll call it a day until I get my hands on the NanoFlash:

I've never shot 24p, I reeeally don't like 24p. Never have. I've been shooting 1080/30p and overall love it except for the one issue. I did some more testing this morning with 1080/60i and it's a little better, but still not crisp with slow movements.

From my eyes, in many circumstances, I feel I could get close enough to, oh, say the F900 to be very happy. The EX1 rocks in most cases. I've read many people say the EX1 intercuts well with the F900 (but I doubt they were motion shots). I'm sure it's not as good as the F900, but I can't backpack and kayak very easily with a F900 so the EX1 looks acceptably amazing. If the Sony was 1/2" CCD chips shooting at 100 mbps with NanoFlash, I'm sure it would get pretty close.

The three options you provided, 1 and 3 are not acceptable/doable. That's just the nature of things, I guess. #2 is my best hope DEPENDING on whether it's skew or codec responsible for 'most' of the motion sharpness reduction. If it's 50/50 responsible then I won't invest in the NanoFlash, the tradeoff for $$ is not there. I can probably buy a Scarlet for that much when it comes out. If it's 90/10, or 80/20 codec driven, then the NanoFlash will be an option because that's a substantial improvement, chip skew be damned.

We'll see when I get to test the NanoPuppy out. Case closed. I do sincerely appreciate everyone's comments.

Hi Buck-
There is no doubt in my mind that the 100 Mbps Long-GOP 1920x1080 MPEG2 CODEC in XDR/nano is above the F900 HDCAM 3:1:1 1440x1080 CODEC. See the attached CODEC chart, created by Sony (we just overlaid the 100 Mbps rate)

Mike, I have no doubt either. BUT the F900 has CCD chips and the EX1 has CMOS. This makes "motion sharpness" the main issue. That's why I keep asking about "motion". No 'stills'. It's a non-issue to me for static shots, the EX1 is amazing. Just comparing "specs" without factoring rendition of motion isn't apples to apples. If the sharpness drop with motion on the EX1 is mostly because of the CMOS chip, it won't matter what kind of codec specs you pour down its neck. If I'm understanding this correctly.

You could have 1000 mbps I-Frame and if the primary cause of the motion sharpness issue is coming directly from CMOS skew upon "movement", then fancy specs bypassing (or increasing) codecs wouldn't be the issue. For other cameras, yes, but specifically for the EX1, not if the limitations are chip driven. That's my dilemma and I'll have to see the results in person. I'm looking forward to it. I *hope* most of this motion sharpness issue is due to 35mbps Long GOP and not the chip itself.

CODEC Quality Chart
 

Sorry, go to the Full Motion Video Comparison thread to view the CODEC Comparison chart.

Hey Mike, where is the Full Motion Video Comparison thread? I just scanned the forum and didn't see it? Sorry.

CODEC Quality Chart
 

Try this link instead: http://convergent-design.fileburst.c...lity Chart.jpg

Buck,

as one of the lurkers and one who does not have an EX1 yet, but has been following this with great interest, please consider reporting back to this thread with your thoughts after you try the NanoFlash! Please.

Cheers, Ian

Thanks! Does this chart making any assumptions on CCD vs. CMOS? I don't see where it does? I could be mis-reading it. If it doesn't, then the "motion" aspect of CMOS isn't factored in. This spec chart might hold up to static shots from CCD and CMOS, and this chart might fall apart upon motion with CMOS. I don't know.

Ian, that is my goal! I'll be a great proponent of NanoFlash if it solves the issue!

Buck,

Ive been following your threads since you first bought your EX1 and understand exactly what you want out of it for those slow horizontal pans of nature scenes. As an EX1 owner I also understand your frustration at not being able to get crisp images of static objects as you pan. I had same problem with some detailed artwork I tried panning across last year. The fine pen drawn lines didnt keep their sharpness even when I panned slowly.

So along comes the Nano Flash to solve this problem but then along comes the fear of Rolling Shutter CMOS distortion potentially counteracting this expensive solution.

I wouldnt worry about the CMOS factor here. As you know the EX1 scans at a constant rate of 60 fields per second from top to bottom of screen. I'm sure for your slow pans this rolling shutter wont have any effect on the sharpness of image because you will probably be moving the camera at only a few degrees a second so any straight verticals such as trees for example should not even have a hint of "fringe blurring" that you might be fearing from CMOS unless those verticals were right up close in the shot in which case they would move across the screen at a faster rate. Even then I dont think it would be a problem.

I say "Fringe Blur" not "Vertical Skew" here because we would be talking about very minute loss of sharp edge rather than full skew from a whip pan.

I think the Nano Flash is just what you need. It should give you the sharpness you want unless your hobby is zooming right into black and white barcodes and panning horizontally.

Another cheaper option you might consider is this.......
Buy a telescope motorized mount for your tripod head and set your camera to interval recording at 1 frame a second and set the telescope head to rotate at 1degree a second. The results will give super sharp images on every frame when you play back. But if you dont want sped up movement of clouds and leaf movement from wind etc I can understand you not wanting to consider that option. I personally like the effect, it makes scenery seem like a minature diaroma somehow.

Anyway, Cant wait to see your results whatever you do.

Kubalsky

Agreed!
Some are speculating way to much that the EX rolling shutter skew is going to add to perceived motion blur. Your fast pan (1/2000 shutter) example observed frame-by-frame certainly looks sharp.

Hi Buck-
No assumption about CCD vs CMOS, the chart just considers the overall quality of the compressed video vs uncompressed (which is the HD-SDI video out).

Numerous tests from Jim Arthurs shows that the XDR/nano 100Mbps 4:2:2 CODEC holds up extremely well, even in CODEC busting rotational tests. Take a look at the following comparison images taking from an EX1 with a 1/2000 shutter speed (to eliminate the blur):

http://convergent-design.fileburst.c...0Mb_Rotate.zip

Mike...

I find it really interesting that compression artifacts in 4:2:0 appear on the macro level as macro-blocking and the micro level as general edge blurring.

The other thing that seems to plague 4:2:0 is intolerance to exposure corrections in post, in the form of macroblocks and grain, especially in the shadows and underexposed areas. Would you expect much improvement in this area with 4:2:2?

edit: In retrospect, the "chroma macroblocks" in shadow areas that are bumped up in post production, may be more related to 8-bit sampling limitations. So the right question would be, will the 10-bit sampling help these very delicate color gradients in the shadows that start to band when exposure levels are raised?

Alexander, thank you so much for that post... it's music to my eyes! It gives me stronger hope that the NanoFlash is my ticket.