M-X Reel

[Jannard]

I saw the film Phillip made online, I downloaded it in 1080p. To me, on on my "27 compu screen it looks rather soft and has moire problems. He did some lovely shots, and has great talent in my opinion, but I don't see how his Skywalker footage shows us anything new about DSLRs.

It also doesn't answer what it would look like recompressed for broadcast, what latitude he would have if he needed to re grade, or how many shots he had to throw out and didn't show us because of even worse moire. That's what I love most about RED vs. other digital means - no nasty surprises and fewer potential mistakes to make on set vs. a cam that has to have its look baked in via a paint and/or will moire or skew without notice.

A generous zone plate gives the 5D about 1.5K of measured resolution...

Jim

 

[Rob Ruffo]

i apologize before hand and i realize this post will be quickly bombed and forgotten but i'm sort of un-impressed with this footage. IMO Laforet's Nocturne footage looked MUCH better. I'm not in anyway a "canon fan-boy" i'm just giving an honest opinion of what i see. There was also grain added to the footage in post which pretty obvious and didn't help with compression artifacts.

Artistically, I can see how someone could be impressed by Nocturne. I am not, but to each their own. But technically you are way, way off the Mark (bad pun intended). My clients would not accept a huge portion of that film on technical grounds. I'm saying re-shoot, at my expense.

The MX footage is clean, clean, clean. They could sell the upgrade for 15K and it would still be a steal. I'm not much of a fan of the MX demo, artistically, either (again, to each their own), but that's really not the point.

 

[Stephen Pruitt]

I am a HUGE fan of the MX video. It's remarkably clean and noise-free. . . the very things I was hoping we'd get from the upgrade.

Stephen

 

[Graeme Nattress]

A generous zone plate gives the 5D about 1.5K of measured resolution...

Jim

As much as I hate to contradict you, I think it's more like 1.4k, and that's generous as it's pretty gnarly

Graeme

 

[Jannard]

As much as I hate to contradict you, I think it's more like 1.4k, and that's generous as it's pretty gnarly

Graeme

Let's use your numbers... you have the charts.

Jim

 

[Graeme Nattress]

With the wind in the right direction, on a static chart which on a long gop codec is giving the image many more bits than it'd get in any kind of real-world footage, and being generous...

Graeme

 

[dino g]

we just shot a lil wayne and eminem video with red#0031 MX enable tonight and last night in very low light downtown, the whole time at 800 asa...the results will speak for themselves.

 

[christopher witzke]

Jim... Has Michael Mann called yet? hehe

 

[Graeme Nattress]

With M-X, I have no problem at all shooting ISO800. The look and feel, even though I say it myself, is lovely.

Graeme

 

[Matthias Martin]

recompressed steering wheel

recompressed steering wheel

i guess i'm missing what people are drooling over with this particular video. i've seen just as good 1080p footage from 5/7/1Ds. not to mention there was aliasing and moire in this video too. (Steering wheel + lines on the audi telephone wires & mercedes)

There is a difference between holding up well enough on the big screen and being rock solid and noisefree.
As mentioned before in this thread, the aliasing on the steering wheel resulted from downscaling and exporting at the same time in FCP AFAICT, silly nearest-neighbor algorithm. It really pays off creating intermediate video files before doing the final encode, this even speeds up the whole process i found.
The proof is in the 4k:
http://www.lightirondigital.com/LightIRON_Digital/LiDigital__MX_4K_REEL_files/R82%204k.0000302.tiff


MX is the seetness BTW. Good to see that REDs sensor program is on such a solid path.

 

[Rob Ruffo]

With the wind in the right direction, on a static chart which on a long gop codec is giving the image many more bits than it'd get in any kind of real-world footage, and being generous...

Graeme

Yes. Try a pan up to a character's face from their hands.. You get Barbie Skin, Blotch Skin, etc. plus... Whatever resolution we do have is not quite real. With missing lines what they are, the whole image is aliasing, at some level. Things like fine textures are symbols of themselves looking somewhat like the real things, rather than true depictions, if that makes any sense.

I look at things closely, and truly love fine subtleties. Not just on screen, but in real life as well. I can easily get an hour's entertainment from the view out my window, watching the sun move over the weathered stone of the old buildings across the street. This is probably why this bugs me so much. Its' images that do not permit one to look closely, negating the essential wonder of HD, let alone 4K.

 

[Rob Ruffo]

I am a HUGE fan of the MX video. It's remarkably clean and noise-free. . . the very things I was hoping we'd get from the upgrade.

Stephen

It's super well done and slick. I'm only saying that the themes are not what I'm personally into. This is a purely subjective comment.

 

[Eric Trageser]

Here is a link to some of the 4K stills for evaluation.

Michael: Thank you for these.

wow. the difference in the cleanliness of blacks / noise floor is huge.

 

[Rob Ruffo]

There is a difference between holding up well enough on the big screen and being rock solid and noisefree.
As mentioned before in this thread, the aliasing on the steering wheel resulted from downscaling and exporting at the same time in FCP AFAICT, silly nearest-neighbor algorithm. It really pays off creating intermediate video files before doing the final encode, this even speeds up the whole process i found.
The proof is in the 4k:
http://www.lightirondigital.com/LightIRON_Digital/LiDigital__MX_4K_REEL_files/R82 4k.0000302.tiff


MX is the seetness BTW. Good to see that REDs sensor program is on such a solid path.

For some reason, FCP, Avid and Premiere do not like doing 2 things at once. Rendering out to a 720p or 1080p (or whatever res you like) uncompressed intermediate and then end-use encoding always seems to give better results.

I sure wish these NLE guys would polish basic functions before adding more gadget-y semi-useless features.

 

[BradWright]

I sure wish these NLE guys would polish basic functions before adding more gadget-y semi-useless features.

Yes, a petition needs to be started to stop the marketing people from demanding worthless features when the core functionality is broken. They need to stop building more crap on top of crap.

 

[Ajit_AP]

Holy smokes! The lowlight capabilities open up so many possibilities.

 

[Michael Hastings]

As much as I hate to contradict you, I think it's more like 1.4k, and that's generous as it's pretty gnarly

Graeme

Okay, this isn't a RED vs. vDSLR question - I'm patiently (well, impatiently) waiting for my Epic-X SN206 and no vDSLR is going to change that - it's a 5d/7D vs. Varicam/HDW-F900 question, because in many ways I trust your data and opinions more than most other sources.

How would 5D (or preferably 7D if you have tested) compare to f900 which I think recorded 1440x1080 or the Varicam which according to one source:

"Panasonic HDX-900, 27H Varicam, HVX-3000 1.1 Mega Pixel imager, frame rates up to 60p; records to DVCPro HD tape at 960x720p resolution 100 Mbs but at 24p really is 40 Mbs?

 

[Graeme Nattress]

Your facts on the Varicam seem correct, although it does tend to look better than the specs suggest. F900 is a decent enough camera too. Remember the 1080p on the vDSLRs measures that same 1440 across.... OK - personal Graeme opinion time - basically, if you want that dreamy everything out of focus no DOF look, there's no advantage to using a proper HD camera over a vDSLR. But if you want things in focus to appear sharp and the right colours, go with a proper HD camera.

Graeme

 

[Michael Hastings]

Your facts on the Varicam seem correct, although it does tend to look better than the specs suggest. F900 is a decent enough camera too. Remember the 1080p on the vDSLRs measures that same 1440 across.... OK - personal Graeme opinion time - basically, if you want that dreamy everything out of focus no DOF look, there's no advantage to using a proper HD camera over a vDSLR. But if you want things in focus to appear sharp and the right colours, go with a proper HD camera.

Graeme

This has to do with some customers that have f900s and Varicams and ask about u/w housings. I have to explain that the starting price for a reasonable size housing (still at least 60% bigger than the R1 housing) is about $15K as opposed to $8K for an R1 housing (and you still need a $20K wide angle b4 lens).

But they can't quite swing the R1 and housing purchase yet - so wondering if I am correct in saying that they can get close to their old performance with a vDSLR - given that underwater subjects tend NOT to push the vDSLR's aliasing issues - for a total of $5 to 6K for camera, lens, and housing.

Of course I'm hoping that once they get hooked they will want to step up to an R1/Epic system.

 

[Rob Ruffo]

Okay, this isn't a RED vs. vDSLR question - I'm patiently (well, impatiently) waiting for my Epic-X SN206 and no vDSLR is going to change that - it's a 5d/7D vs. Varicam/HDW-F900 question, because in many ways I trust your data and opinions more than most other sources.

How would 5D (or preferably 7D if you have tested) compare to f900 which I think recorded 1440x1080 or the Varicam which according to one source:

"Panasonic HDX-900, 27H Varicam, HVX-3000 1.1 Mega Pixel imager, frame rates up to 60p; records to DVCPro HD tape at 960x720p resolution 100 Mbs but at 24p really is 40 Mbs?

I'd say an Ex1 will give you FAR better results than that Varicam, under almost all circumstances. Sharper image, better codec, better everything. An Ex3 with a better lens than the kit - you're flying. Get a $3K nanoflash and then you are getting comparable results than an F900, lens and DOF issues excluded.

An F900 has lovely contrast, lovely color definition 9it's really the color def that amazes on this cam, I.M.O.), lovely many things, but looks ever so slightly less sharp than an Ex1/ex3 (particularly with nanoflash). As such, reframing for stabilization can get tricky.

Put a still frame from any pro or prosumer vid cam next to a 5D 7D still, and you really do see that the 5D has weird textures, is soft, has a crappy codec, etc. Vimeo type sites, with highly compressed 720p are highly misleading.

On a 1080p screen, the difference is startling.

The biggest problem, again, is moire. These Vimeo and other website posters are not doing pro jobs, for the most part, they're just fooling around. If they are pro jobs, the standards are not so stringent, as their buyer is paying low dollar, so is willing to look aside from a certain number of problems. It's fine if they have to throw out 60% of their footage. In my world, and I would assume yours, such a situation would be dire indeed. I'm sure there are exceptions to this, but not so many. People often state the very few ads shot on 5D, as though surely anyone shooting an ad must know what they are doing and have very high standards. This is not true. Some people don't dot any of their 'I's, or are grossly incompetent even, and work anyway. It's a bit like saying that if you are a Hollywood screenwriter you must only write good scripts, or that because you saw a major studio film with wooden, empty acting, that now poor acting is OK and you needn't bother with trying to get a better performance.



I work like a dog for my shoots, as I'm sure do most of you here, sometimes seriously risking my safety and my health to get a shot. I feel that putting all that blood sweat and tears behind very unreliable gear would be rather masochistic. I mean 30K for a Red is teh cost of a midrange car. Given the choice, I'll keep my old car and save myself 5D grief.

That said, if you are mastering in SD, do not plan on grading, and if you do not plan to shoot any buildings in focus, the 5D can be a great tool.