Full-frontal insidious attack on Red, by indie director Rian Johnson (2008)

[gbalaji]

Dan,

Is it possible for you to arrange for some frames. In India, there are whole lot of camermans shooting on Red for the sake of doing it. For the project I work as DIT / DAM /DAS, I insist them to understand the limitations of RED and do a test pring. But DOPs are not interested on that and they are very much worried about the end result.

In one of my project, I got complain from post department that the footage looks too soft and after searching entire reduser forum, I found that rendering DPX from RedCine will be softer and needs to add little bit sharper in post where as RedAlert has built-in sharpness.

My only concern about RedBull post is, Is it fake or there is a real person. Unable to find even single email address or Phone no from their website. I don't go by IMDB alone without checking the real facts.

 

[Dominic Cochran]

He links to his e-mail at the bottom of the page where he says he wants to hear if there's evidence contrary to what he's seen.

 

[Andrew Walker]

Hey Red, can I get a refund? I saw BRICK, and i trust this guy........

Hahahaha. As always Richard the comedy is welcome.

This is something as a Red user that we all have to deal with. People that have no idea of what they are talking about. Or even worse they heard from someone that heard from someone else that they had problems with the camera.

 

[Joe G.]

-

-

Can we get the specifics of how he got those results? Each step should be analyzed, starting with the build, the specific areas he saw "clumping" and the codecs and software used in post.

If it turns out there were compression errors, or dark "noise", then we may be able to HELP him smooth the footage out. Perhaps he will even show some original RAW files where the clumping supposedly originated. That would go a lot farther than just bickering back and forth on message boards. Someone from RED should contact him immediately with an offer of helping clean up his footage with their expert analysis.

My 2 cents.

 

[JohnLands]

Red should just rent a theater in los angeles and show a 20-30 minute demo of clips that run over and over again and invite the public to come view and see for themselfs. This would save alot of productions alot of money time and effort. And sell the cameras to alot of people that have only heard or seen bad things like the person posted above. Have a q and A from red after each presentation and invite all other companys, sony panasonic....... to do the same thing.

 

[reality]

It's almost the same as what I've heard out of Panavision and Sony employees so I wouldn't doubt that it's roots are corporate inspired FUD and propaganda.

 

[Esteban Sosnitsky]

PLEASE SOMEONE CORRECT OR INFORM ABOUT THIS:

On this article it says:
"RED's sensor does indeed have 4096 photosites across its imaging area, one for each pixel, but here's the rub: each pixel in a digital image is made up of three channels: red, green, and blue. The Red camera's image sensor has 4096 TOTAL photosites across the image area, NOT 4096 for red and 4096 for green and 4096 for blue (you would need 12288 photosites for a true, uncompromised, 4k image). So, Red camera is using 4096 photosites to gather a supposed 12288 pieces of data. "

I thought the camera was 4.5K Giving more room for the "missing pixels", that would be a lot more than 4096 right? The interpolation does occur but you can easily imply what the next pixel is thanks to the Bayer mask. RIGHT? I thought the real resolution was between 2.5 to 3.5 depending on project seetings but less than an F900? PLUS he claims that you should multiply the native resolution of the high end HD cams with 3 ccd's. BUT I thought you needed just like in the RED 3 pixels to figure out what the color is, so its 3x1=3 RIGHT?? PLUS the 3CCD sensor high end cameras (all except the f35) are 2/3 inches, so when using a cine lens you are compressing the image quite much !! RIGHT??? When this guy talks about compression he is also stating his conclusions based on file sizes??? He says that the file size is huge in red because its inflated, but isnt it normal when you have a RAW file?? The raw file has not a lot of information implied or attached to that file itself so of course an F23 file would be bigger, but its not RAW! Its like comparing a film negative scanning it a different sizes! BUT PLEASE DO CORRECT ME IF AM WRONG!!

Even though doubts may arise I have seen the results on the RED tent at NAB, I have seen the results on SD CRT monitors, side by side comparisons with the high end HDTV I dont need to do a printout, I have done scaling in SD to know that if you have an HD image there is nothing you can do to get a bigger image than that no matter what interpolation you do! SO could someone explain me what is he talking about? thanks.

 

[Esteban Sosnitsky]

Red should just rent a theater in los angeles and show a 20-30 minute demo of clips that run over and over again and invite the public to come view and see for themselfs. This would save alot of productions alot of money time and effort. And sell the cameras to alot of people that have only heard or seen bad things like the person posted above. Have a q and A from red after each presentation and invite all other companys, sony panasonic....... to do the same thing.

Only in LA????
Man, that would require like 14 plane hours for me!!

 

[Johann Schulz]

I think he may have been watching footage of that codec-killing tree stump from a few months back?...

 

[rileytaylor]

He may be full of crap, but his music video is amazing. http://www.rcjohnso.com/goats.html

 

[laguun]

Last month alone we did >250 film copies (arrilaser 4K/baselight).
They were excellent.

I dont know what the poor director was doing wrong, but for sure he doesnt understand todays image acquisition technology, and he also has difficulties reading -the red has 4900, not 4096 photosites, of which 4520 are used to debayer an image.

Furthermore in his compression calculations he is misplacing CMOS RAW and RGB - pretty incompetent for 2008, as these technology is common in photography since 2002.

The images a competent director/dp/post team can produce with red are simply awesome.

 

[BobSimon]

First, here's the link to the article, which I don't think has been posted: http://www.rcjohnso.com/REDBULL.html

Second, my take on this: I've heard Johnson talk, interviewed on NPR's The Treatment, and he sounds like a really common sense kind of guy. That said, he's one of the luckiest newbie filmmakers in the last decade with a first feature out of the gate that is actually amazingly well shot on 35mm, and somehow pulled off for under $500K. (Brick.) Several shots in that movie were jaw dropping to anyone who knows what a $500K budget means.

When I saw Red project the 4K "milk girls" footage at the Nuart theater in West LA about 2 years ago both compressed and uncompressed, that was what sold me. No one could tell the difference. I've since seen both NAB screeners and in both I'd say 90% looked as good as any film I've seen projected, and the rest showed the limitations of the medium. That's a better ratio than any Varicam or F900 reel uprezzed and then projected at 4K would ever be.

On the issue of compression, people should brag about the 36x compression! Every few years there's a new breakthrough in compression algorithms and Redcode has been a major breakthrough, particularly with its inclusion of RAW data. H264 is about 8 times more compressed than DV, but having come along 15 years later produces an image WAY better than DV.

I don't think Rian Johnson needs to be attacked on this forum or anywhere, but I do hope to see how he responds to information he might not have had at the time he and his DPs did their testing.

RED is cheaper than F23 or D21. So, people who cannot afford F23/D21 can buy a RED and shoot a film. That is good. The main thing is the price. That is revolutionary in my opinion.

How can you make a feature movie on 35mm film for $500K? That is just the cost of fim stock, right?

 

[Dan Hudgins]

In the right hands...

In the right hands...

The images a competent director/dp/post team can produce with red are simply awesome.

I would have to agree with that, when exposed properly with the right lighting ratio the RED ONE (tm) camera seems to make images that are overall better than 35mm film.

The RED ONE (tm) is not a "magic box" that you can point at anything and get eye poping results. It is a sort of RAW mode camera, and needs someone who knows what they are doing behind it, and setting up the lights.

If you want to see how far you can push the limits of its range, you are going to see bad results, but if you shoot film at three foot candles is it going to look better? Not from anything I have scanned off film do I see that being the case?

If there were no compression you would see sensor noise also, just like you will see grain in film when you under expose too much. Some cameras may use more DSP to hide those issues, but a sort of RAW camera lets you leave that to post, if it is in the camera you cannot adjust image processing later.

I think the RED TEAM (tm) should make a 35mm 4K 2000 foot film out demo real and rent it out so people can see what the camera can do when used under the best and not worst conditions, then it is the users fault and not the camera makers if the footage looks like crap...

==

Having done "RAW" film scans, I can say that most of what people think film looks like seems to be heavy post processing with "cleaning software" of the film scans, rather than the data coming off the film. Maybe RED TEAM (tm) should look more into what is used to process film scans and how that workflow might be used to impact some problems with the RED ONE (tm) camera. I do think that they need to explore the OLPF vs. chroma moire better, and look at some of their chroma moire filters to imporve the results. Also people should not edit using double compressed formats, and think the artifacts are in the camera images, if you do not see the artifacts in the 16bit TIF from REDCINE (tm) then you probably put them in with something else along the workflow, before people say the camera is bad they should look at the source of the artifacts since the de-Bayer (settings) and re-compression for editing and filmout not from 16 bit TIF makes other problems. DPX do not have enough bits to be CC after being output, you will get gaps in the histograms, uncompressed 16 bit TIF output is the minimum option to hold all the image tones through heavy CC.

 

[Dominic Cochran]

How can you make a feature movie on 35mm film for $500K? That is just the cost of fim stock, right?

Sweet jesus no. Film isn't THAT expensive.

 

[Graeme Nattress]

Tis rather funny indeed. It's like saying a 12mp DSLR is crap and you'd get a better still image from an F900, F23, F35 or Genesis. Like what?

As for compression, we take 4096x2304*12bits = 14155776 bytes, or 13.5MB, down to 1MB or 1.2MB or in that region, giving around 12:1 compression, give or take. Remembering of course, that the modern wavelet compression use is a lot more efficient than the block based codecs used by other cameras.

And of course, resolution is not in the photosites, but actually measured using charts, counting lines visible before onset of excessive aliasing.

The facts are that the measured actual real world resolution, measured, from the RED one is greater than that of any HD camera.

And anyone who's seen the RED 4k projected at NAB, or IBC, or at the RED office can confirm what the image looks like.

The problem is not that this guy doesn't understand all the nuances of how cameras work, but that he appears not to want to understand. He's obviously seen some bad footage, and we all have, from all cameras.

I've written more posts on bayer patterns and how they work than practically anything else. It's appear he's not read them, but you, my reader probably have, and that makes you the knowledgeable one.

I'm just please I wasn't drinking my coffee when I read this, or I'd have an awful mess on my desk and monitors to clean up.

Graeme

 

[J.R. Hud]

That was a very cool video.

 

[Jeff Coatney]

I encounter people with a negative view of the Red camera daily. Without fail, the loudest, most vocal critics have never seen Red footage projected in a screening room. One of the most dismissive people I've met regarding Red, rented our camera after I showed him the footage. I actually enjoy meeting people who have not seen the footage, and feel free to bash it anyway. I always ask if they've seen Red footage, and the answer is almost always no. They quickly change their tune after viewing the footage. And the people who filled their heads with that non-sense quickly lose credibility in their eyes.

Judge on results. See the footage.

 

[Lewis-M Soucy]

He may be full of crap, but his music video is amazing. http://www.rcjohnso.com/goats.html

I don't know really, because "it's seductive to see beautiful footage played as a compressed and down-sized Quicktime file on your computer and be wowed by it"...



Just kidding, he truly is talented. (and the concept of that music video would make a great HP commercial!)... Kiddin again, he is talented...

 

[Thor Wixom]

Brick was one of my favorite movies of the last few years...

...but being an exceptional artist does not make you a technician.

-Thor

 

[Daniel Browning]

-Is Red One's compression really 36 to 1?!?

When it comes to image quality, any talk of the compression ratio is disingenuous. I have mentioned this before. The upstream signal, codec, and encoder are what actually determine the quality.

It's like evaluating engine power based on its sound level. Sure, louder engines are generally more powerful, but not always (as proven by the kids with with those lame mufflers).

Another example is talking about "how much it costs to fill up" when evaluating fuel efficiency. Obviously, metrics that fluctuate with tank size, fuel price, and amount of gas in the tank do not, by themselves, indicate efficiency.