Welcome to our community

Be a part of something great, join today!

Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

Jumping off the Red Wagon

Status
Not open for further replies.
4x more data.

Yet another her reason to stick with RAW but prores is obviously the king. Love it as a delivery codec prefer r3d for acquisition.
 
Jarred basically confirmed 4k prores the other day, just still waiting on apple approval
 
Bob Gundu said:
Still scratching my head over not having internal 4K prores or at the very least 4K output on the weapon.
Click to expand...

I'm sure you know this already, but 6k R3D files are way smaller in data than 4k PRORES of equal dimensions. And not colour backed like PRORES either.

You can drop a RAW directly into a time line and you can grade RAW in that time line...

Not sure what the attraction to PRORES is anymore. ? other than low res editing ?

I only use PRORES when Im forced to edit swiftly on a laptop. With bigger machines I don't go near PRORES. Your limited within that colour baked world and your images then become compromised in that world. Not to mention your hard drive has to be twice the size to handle 4444 4k even 422HQ 4k is bigger than R3D's.

However, I do use PRORES for final Mastering once I get that perfect look Im after.

Food for thought.
 
Has anyone read the result of Geoff Boyles test at the University of the West of England. This is not meant to challenge anything or change anyones opinions, I'm just curious about what people think of it.

According to the test....


Red Dragon with the skin tone filter in daylight showed a D.R of 12.5 stops

Alexa showed a D.R in daylight of 15 stops

The Varicam in daylight 12.5 stops

The C500 in daylight 12 stops



I'm not trying to instigate another Reduser "battle"... I'm just wondering if anyone has seem the test and what they think of it.


Here is a link to the results below

http://www.cinematography.net/CML-UWE-tech.html
 
Mark Toia said:
I'm sure you know this already, but 6k R3D files are way smaller in data than 4k PRORES of equal dimensions. And not colour backed like PRORES either.

You can drop a RAW directly into a time line and you can grade RAW in that time line...

Not sure what the attraction to PRORES is anymore. ? other than low res editing ?

I only use PRORES when Im forced to edit swiftly on a laptop. With bigger machines I don't go near PRORES. Your limited within that colour baked world and your images then become compromised in that world. Not to mention your hard drive has to be twice the size to handle 4444 4k even 422HQ 4k is bigger than R3D's.

However, I do use PRORES for final Mastering once I get that perfect look Im after.

Food for thought.
Click to expand...

Yes great point Mark. I'm not a proRes editor either and I think it would at least get newbies into the RED world, and I'm sure it wouldnt take long for them to go the Redcode route too. It may be the Trojan that RED needs.
 
Nosa, I saw Geoff's tests, and took them with a grain of salt, for various reasons:

1) Geoff's determination of DR was apparantly decided by his subjective choice of acceptable/usable noise in the blacks/highlights.

As many redusers are aware, the amount of "noise" in a RED capture is highly influenced by your signal chain, so:

2) Geoff' insistance on handicapping the Dragon and shooting it at 5K (to keep it at s35), clearly had an influence on the signal chain - 5K is less clean than 6K, as you are punched in on the noise (notice also that Geoff I believe did not similarly punish the BMCC for being s16).

3) Compression. I don't believe we know what compression he shot. I never saw it listed. Was it 5K 5:1, 5K 8:1, or 5K 10:1. These choices have a major outcome on noise, and Geoff's choice of compression was not clearly labeled the few times I checked. Like 5K, the compression appeared to be a personal choice on his part.

4) Blackshading has a major influence on noise. To not do so before a test, could dramatically affect noise - Dragon is much more sensitive to MX in this regard. Again, we have no idea if Geoff properly blackshaded before his test. Each time I checked, this was never made clear.

My point being - Dragon shoot 5K @ 8:1 w/out a blackshade will have significantly more noise than Dragon 6K 5:1 with a proper blackshade. Without more clear info, we have no idea if were seeing the results of the former or closer latter (ie at 5K) in Geoff's test. Geoff's failure to mention compression or blackshading makes me assume he isn't aware of their significance. I dont believe he's a regular RED shooter anymore, so this is actually understandable.

In fact, if this is true, this says a lot about RED and how it lives within an Alexa dominated marketplace. Geoff, a seamingly regular Arri shooter, is probably used to turning on the Alexa and it just "works" - no blackshading, minimal recording options (RAW or a few standars flavors of Prores - it probably generally stays on Prores HQ 4444?).

When faced with a Dragon, Geoff (or similar Alexa users) may not intuitively realize there a number of more custom RED specific variables to consider:

A) which OLPF?
B) Have you blackshaded?
C) What resolution? 6K? 5K? 4K?
D) What compression? 5:1? 8:1? 10:1?

Each one of these RED specific choices has a major affect on your signal chain, outside of the more universal choice of EXPOSURE.

Not that I think REDs level of choices are bad - in fact when harnessed they are GOOD (and give you immense control over your capture). But they are certainly a complex set of parameters that demand time and experience to process. Weapon is addressing some of these parameters in interesting ways (auto blackshading could be a major timesaver and gaurantee of optimal capture).

That said....this is a long way of saying...without knowing more about Geoff's signal chain, I've taken his results with a pinch of salt.

Ironically, if asked to reveal his signal chain, my guess is Geoff would likely blame Dragon for being too complicated in the first place. Not fair, but this seems to be a perception from the more streamlined Alexa POV, imho.
 
Last edited:
Nick Morrison said:
I saw Geoff's tests, and took them with a grain of salt, for various reasons:

1) Geoff's determination of DR was apparantly decided by his subjective choice of acceptable/usable noise in the blacks/highlights.

As many redusers are aware, the amount of "noise" in a RED capture is highly influenced by your signal chain, so:

2) Geoff' insistance on handicapping the Dragon and shooting it at 5K (to keep it at s35), clearly had an influence on the signal chain - 5K is less clean than 6K, as you are punched in on the noise (notice also that Geoff I believe did not similarly punish the BMCC for being s16).

3) Compression. I don't believe we know what compression he shot. I never saw it listed. Was it 5K 5:1, 5K 8:1, or 5K 10:1. These choices have a major outcome on noise, and Geoff's choice of compression was not clearly labeled the few times I checked. Like 5K, the compression appeared to be a personal choice on his part.

4) Blackshading has a major influence on noise. To not do so before a test, could dramatically affect noise - Dragon is much more sensitive to MX in this regard. Again, we have no idea if Geoff properly blackshaded before his test. Each time I checked, this was never made clear.

My point being - Dragon shoot 5K @ 8:1 w/out a blackshade will have significantly more noise than Dragon 6K 5:1 with a proper blackshade. Without more clear info, er have no idea if were seeing the results of the former or latter (even at 5K) in Geoff's test. Geoff's failure to mention compression or blackshading made me assume he wasn't aware of their significance. I dont believe he's a regular RED shooter anymore, so this is actually understandable.

Nevetheless, because of the many uknown variables, no matter how well intentioned, Inhave taken these tests with a grain of salt.
Click to expand...

Compression was 5:1 - noted on the Dragon Skin page.

I couldn't find anything on the black shade - I suppose it is possible that Geoff - who by any measure is very experienced - could have made a mistake... but I doubt it. I also think he could have squeezed a bit more out of ISO 250, but only a little. And looking at these clips in resolve, I don't think 5k versus 6k would really move the needle. There is a lot more noise in the Dragon image.

I think the test pretty well spells it out. It is what it is. Dragon is noisier in all channels, especially the red channel - Dragon also clips sooner than the Alexa by a couple stops.

That's just how the cameras behave. I don't think there is much wiggle room in these tests unless they didn't black shade or they black shaded incorrectly. Maybe Geoff will chime in on that.
 
Nick,

your post underlines the very things many people don't want to go to the trouble of knowing or understanding. Black shading, compression, operating temp, cropping factors "most" folks don't give a shit. They want to turn the effing camera on and shoot.

The boot time has crept up to 32 seconds last time I checked. So I know enough to put the battery on and turn on the camera before I add the lcd, lens, rods, mb etc but many other cameras are instant on.

Its the small, seemingly insignificant things that tend to matter. Personally, The geek in me loves and appreciates this stuff and my curious mind and problem solving nature is ok with having to learn, know and understand this stuff.

Others, not so much, so they lean toward what is good and what is easiest.
 
David Battistella said:
Nick,

your post underlines the very things many people don't want to go to the trouble of knowing or understanding. Black shading, compression, operating temp, cropping factors "most" folks don't give a shit. They want to turn the effing camera on and shoot.

The boot time has crept up to 32 seconds last time I checked. So I know enough to put the battery on and turn on the camera before I add the lcd, lens, rods, mb etc but many other cameras are instant on.

Its the small, seemingly insignificant things that tend to matter. Personally, The geek in me loves and appreciates this stuff and my curious mind and problem solving nature is ok with having to learn, know and understand this stuff.

Others, not so much, so they lean toward what is good and what is easiest.
Click to expand...



People always want one less thing to worry about.... all the stuff above is just more stuff you have to deal with... inevitably it slows you down and leaves more room for error. On the other side of the argument, it gives a knowledgeable D.P or operator finer control of the tool.
 
NOSA OBAYIUWANA said:
Has anyone read the result of Geoff Boyles test at the University of the West of England. This is not meant to challenge anything or change anyones opinions, I'm just curious about what people think of it.

According to the test....


Red Dragon with the skin tone filter in daylight showed a D.R of 12.5 stops

Alexa showed a D.R in daylight of 15 stops

The Varicam in daylight 12.5 stops

The C500 in daylight 12 stops



I'm not trying to instigate another Reduser "battle"... I'm just wondering if anyone has seem the test and what they think of it.


Here is a link to the results below

http://www.cinematography.net/CML-UWE-tech.html
Click to expand...

Here's the problem with that test: "0 noise base, very noisy as measured but visually less so"

Many people just don't know what to do with DRAGON's noise character. Some, like Phil Holland, Mark Toia, Gunleik Groven, and others, make it look beautiful, natural, and organic. Others make it look terrible, degraded, and sloppy. In this test, they just dropped back and punted the whole bottom range of the sensor, which misrepresents what the DRAGON can do, and especially what the DRAGON can do in good hands. I trust that the GIO scope can measure 16 stops of DR, and I concede that what gets binned into 1 and 16 is pretty much crushed black and clipped white. It is not dreadfully impossible to get 14 meaningful stops between those extremes using simple temporal NR and/or D.E.B. for the blacks and to let DRAGON do what it does for the whites.

A proper blackshade is obviously crucial to get good shadow performance from DRAGON.

I think RED may have over-estimated how much effort its customers were willing to do in the name of ultimate image quality and fidelity. If we could have had 3 stops less shadow noise at the expense of 1/2 stop of DR, that would have cost it some points with the top 1%, but it would have made the remaining 99% much happier.
 
NOSA OBAYIUWANA said:
On the other side of the argument, it gives a knowledgeable D.P or operator finer control of the tool.
Click to expand...

So true. So many people who take the time really squeeze amazing images from this camera and many of them are posting in this very thread.

To to me it's about raw and resolution, I do think that those two factors are cornerstones of lasting images (even in digital). My Epic dragon is by far the best camera I have ever owned and I feel they are just the finest tool around for making images for me personally.

i respect all the people out there making their images and I respect their right to choose whatever tool they want. But I do roll my eyes at some of the stuff I have seen and heard over the years about "issues" with RED.
 
Mark Wilkinson said:
Compression was 5:1 - noted on the Dragon Skin page.

I couldn't find anything on the black shade - I suppose it is possible that Geoff - who by any measure is very experienced - could have made a mistake... but I doubt it. I also think he could have squeezed a bit more out of ISO 250, but only a little. And looking at these clips in resolve, I don't think 5k versus 6k would really move the needle. There is a lot more noise in the Dragon image.

I think the test pretty well spells it out. It is what it is. Dragon is noisier in all channels, especially the red channel - Dragon also clips sooner than the Alexa by a couple stops.

That's just how the cameras behave. I don't think there is much wiggle room in these tests unless they didn't black shade or they black shaded incorrectly. Maybe Geoff will chime in on that.
Click to expand...

Mark, 5:1 is good to hear, Im happy to stand corrected. Nevertheless:

1) 5K is still 16.7% more noise, mathematically speaking, than 6K - and while not a major difference in the real world (I agree), when pushed to extremes as Geoff was doing, could in theory tip an extra "point" of DR here or there, esp if Geoff was being especially demanding (as he appears to have been, and rightly so).

2) Would be good to know if cams were blackshaded, yes!

4) One thing his test didnt emphasize is not all captures were created equal - the Alexa was either 1080 or 3K, while Dragon was 5K (not even the 6K it was capable of). Depending on your application, this could be a big deal (ie for VFX, 4K delivery, etc).

5) Finally, for a different take on Dragon vs Alexa, look at this test from Travis.

At 800iso at a 5.6, Travis found Dragon 6K downscaled to UHD sharper and LESS NOISY than Amira upscaled to UHD.

Having seen Geoff's tests, I would have thought this unlikely.

But in Geoffs defense, his test wasn't about BEST image at optimal exposures...it was about DR under extreme conditions....something Alexa clearly excells at (to this day).
 
Last edited:
And Alexa could've been 3k.

I don't want to go into discussion on Geoffs results, but I think it's silly to suggest that he does anything to deliberately cripple anything, and his reasons for shooting 5k are good.
And many would agree. That if you know a lens range and change the FOV you no longer know that lens-range. (or the quality of the lens if you shoot it wider than intended).

Also, knowing that a lot of TV is shot 4kHD, even on Dragon, the case that Dragon CAN be shot at 6k, is not really the most realistic scenario.

Geoffs tests are as far as I know good, with a very practical approach and seriously and well documentedly done.
And even though I agree that they should be read like any bible, the good part with them is that they are done in a way that you or I could recreate.
They're actually well documented.

Which basically means. if u do what he does, you'll probably get similar results.

And I like that a lot about them.

Now DxO also do their tests in a good manner.

But I don't think those results nullify eachother.
 
Gunleik Groven said:
And Alexa could've been 3k.

I don't want to go into discussion on Geoffs results, but I think it's silly to suggest that he does anything to deliberately cripple anything, and his reasons for shooting 5k are good.
And many would agree. That if you know a lens range and change the FOV you no longer know that lens-range. (or the quality of the lens if you shoot it wider than intended).

Also, knowing that a lot of TV is shot 4kHD, even on Dragon, the case that Dragon CAN be shot at 6k, is not really the most realistic scenario.

Geoffs tests are as far as I know good, with a very practical approach and seriously and well documentedly done.
And even though I agree that they should be read like any bible, the good part with them is that they are done in a way that you or I could recreate.
They're actually well documented.

Which basically means. if u do what he does, you'll probably get similar results.

And I like that a lot about them.

Now DxO also do their tests in a good manner.

But I don't think those results nullify eachother.
Click to expand...

Fair enough, Gunleik. Ive updated my post accordingly. Thanks for being a voice of reason - much appreciated!
 
I think we should organise our own OBJECTIVE test.

I've been thinking of doing this for sometime now...

...anyone interested in doing this in the autumn @ the Sony centre in Pinewood studios?


Possible cameras available

Red Dragon (or Weapon if it is shipping by then)

Sony F55

Sony F65

Alexa XT (or SXT if it is available by then)

Arri Amira

BlackMagic 4.6K (not the mini)

Canon C300 mark 2

Panavision Varicam ...don't have access to one but I may be able to ask the guys at CVP to loan us one a the day...or two
 
Status
Not open for further replies.
Back
Top