We've relaunched as a full community! Get the scoop:

November 2, 2012

Shoot 8K on the Sony F65 with a Future Firmware Update, Plus More Details on Sony's XAVC Codec

As you likely already know, Sony has made some pretty major announcements recently, which we've been hard at work following, though pricing details are currently still under-wraps (likely because of a major price cut by RED). We do know that the F55 will be capable of on board 4K compressed shooting and 4K RAW with an external recorder, while that recorder will also enable 4K RAW capabilities with the on board 2K-capable F5, and bring a 2K RAW recording option to the FS700. Plus, further RAW resolution abilities are down the road for the F65, to be activated in conjunction with an upcoming firmware update. As far as handling some of this media goes, Sony has developed new AXSM media cards (for use in the external recorder module) as well as new XAVC codec technology for 4K shooting (among other things) -- a lot to take in all told, we know. With all these details (and X's, and A's) flying around, it can be tough to see the importance of each one. Rest assured, though, that this new codec marks an important new step for Sony and its shooters. Read on to find out why, with some discussion of the F65's future as well.

The element that ties these new cameras and recording capabilities together -- via the SxS PRO+ card slots native to both the F5 and F55, as well as the AXSM (512GB) card-accepting AXS-R5 RAW recorder -- is XAVC, a robust and flexible format. According to Sony, the development of this codec highlights the company's "commitment to bringing high quality content to the consumer market," because the "quality of HDTV programming originated in 4K is simply stunning and will allow both broadcasters and the production community to build a future-proof catalogue of high-end content, now." Let's see, now: future-proofing, and at 4K? -- this sort of terminology may ring a few bells for some of us, but that doesn't mean Sony isn't correct about this.

XAVC will support everything from 4K to proxy resolutions (with "1080 50/60P infrastructure capability") and both intra-frame and long GOP compression schemes. It is built upon MPEG-4 AVC/h.264 level 5.2, which Sony says is "the highest picture resolution and frame rate video compression codec based on industry standards." The format will also be capable of 12-bit, 10-bit, and 8-bit color depths and 4:4:4, 4:2:2, and 4:2:0 color-sampling resolutions. The MXF wrapping container is supported, for those worried about that. While XAVC has the potential to do all of this, Sony says that it will be up to "each manufacturer [to] decide which profile and operating point of XAVC they will implement in their products."

Here are the industry leaders and developers already planning to support, implement, and/or contribute to the XAVC format and its workflow (via Sony):

  • Non-Linear Editors: Adobe CS6 with Rovi Total Code Plug-in installed, Avid, Final Cut Pro X, Grass Valley, Quantel, Rovi, and Sony Vegas Pro 12
  • On-set dailies: Assimilate, Codex, Colorfront, FilmLight, MTI Film, and YoYotta
  • Color Grading: Assimilate, FilmLight, and Quantel
  • Software codec: Rovi MainConcept SDK
  • Codec board: Matrox

The SxS cards we're all used to can be used to shoot XAVC in 2K up to 30 fps -- beyond this, the new SxS PRO+ cards are necessary. The bottom line for some of us at this point may simply be that the XAVC format allows on board 2K shooting for the F5 at twice the bit rate even of Sony's newly-50Mbps MPEG-2 HD recording mode (which itself has come a little late to the party). Or, the bottom line may be that XAVC offers a smaller-footprint alternative to the less-compressed and higher bit rate-capable SR codec (a more storage intensive HD-res-only option), which will apparently be migrating from the tape-based HDCAM-SR system to the fully-tapeless digital realm.

The other major update for Sony regarding XAVC here is that it's being developed as an open format, not entirely unlike CinemaDNG (though note, the latter is uncompressed). Sony says this allows for "a license program for other manufacturers in the broadcast and production industry to develop their own high quality and high frame rate products." This could prove to be really major, because in a way it brings a future-proof and modular, evolution-capable mindset to the format itself. The software can become more advanced parallel to new hardware, as long as updated standards are accepted and support is offered for farther down the workflow pipeline -- and ideally, later development of the codec would not rely on any single given company, though there are no guarantees.

Some of us may be wondering, how exactly do XAVC and the AXS-R5 affect the F65s already out working in the field? While details regarding this matter are scarce (and Sony hasn't included much material regarding it, considering all the data included in recent press releases), Studio Daily has mentioned that Sony should have a firmware update ready for next year's NAB that "will enable 8K x 4K, 6K x 3K, and 8K x 2K" recording abilities for the F65's much-lauded 8K sensor. If you're wondering what application shooting in 8K x 2K would have (and it would be understandable to wonder such a thing), Sony is apparently working with Zeiss on anamorphic options that would be revolutionary in the digital realm. The company stated at an October press conference: "This anamorphic experience is not stretching or diluting information. It's really taking full advantage of the 8K sensor in the F65."

While it's possible that the XAVC codec could find its way to the F65, none of the publicized materials suggest the AXS-R5 recorder is 8K-capable. The question is, do existing recorder options (like the SR-R4) for the F65 already have the innate ability to record full-res, which would then be enabled when this firmware (perhaps along with some related updates) is released, like waking some sleeping giant? The matter may require some additional research, because we've always known the F65 would achieve greater powers down the line, we just didn't necessarily know it would work backwards, too -- and Sony has also stated that "from the F65 RAW we will be able to get a minimum of 6K, even with RAW recorded today," and that there are "many exciting updates for the F65 Digital Motion Picture camera, including 6K and 8K de-mosaic by way of updating our SDK!"

My understanding of what this means is, in the future, greater resolution from legacy F65 footage can be obtained by retroactively re-de-mosaic-ing that old footage with a new algorithm -- an ability likely to-be bundled with the major firmware upgrade. So... will we be hearing another announcement for another recorder, this time for the F65, when the full firmware update specs come out? Maybe not, but we can only guess, because again, details are thin at the moment. However, if Sony has already been implementing the forward-looking mentality demonstrated by what they've most recently announced, it's not outrageous to assume the AXS-R5 may be compatible with the F65 to some capacity in the future -- though we think this is more unlikely than not. The aforementioned 8K-enabling firmware update will come at some cost, also as of yet undisclosed.

In the meantime, here's a video demonstrating, among other things, the very high native dynamic range of the F65, also via Cinescopophilia:

Do you guys think Sony's really stepping up to the plate with all this new tech? Do you think with all these details working together, that Sony could carve out an even more significant professional (and prosumer) marketshare?

Links:

[via Cinescopophilia]

Your Comment

65 Comments

I'm not sure how exactly they propose to be able to record 8K from a sensor that isn't 8K.
Basically, their sensor is a 6K Bayer pattern sensor with the pattern and pixels rotated 45 degrees, and then they count by stair-stepping two overlapping lines of pixels for the horizontal number - which to me sounds like PR and nothing more. 6k by 3k would have been great for marketing in a 4k world.
Since you cannot save a computer image into an 8k by 2k image (seeing as a computer's pixels are not rotated 45 degrees and don't overlap diagonally) shooting 8K will always need to be resampled into a smaller container, no matter how you spin it.

This article explains it pretty well, using diagrams to demonstrate the method with which they can claim 8K:
http://camerarentalz.com/sony-f65-6k-sensor/

November 2, 2012

0
Reply
Thor

PS, just to be clear, the F65 shoots great images. They're just not 8K, nor will they ever be, no matter what Sony's marketing department says.

November 2, 2012

0
Reply
Thor

I did not know that :) Thank you for the info, I'll definitely be looking into that.

November 2, 2012

0
Reply

That article has its own fuzzy math to prove a point.

I don't get where the dude came up with 6k based on the ratio.

8K could also mean 7680×4320 which is multiples of 1920x1080

http://en.wikipedia.org/wiki/Ultra-high-definition_television

It's no different with Red One and some detractors saying its not true 4k.

November 3, 2012

0
Reply
Thomas

I mean the 8k output not sensor could be 7680 × 4320

November 3, 2012

0
Reply
Thomas

In TVLand 7680 × 4320 might be considered 8K, but in the real world, 1K is 1000 pixels (sometimes 1024 pixels) but never 960 pixels. So in reality, 7680 is 7.6K and not 8K - which is irrelevant anyway because this sensor is not 7680 × 4320 by any stretch of the imagination. You'd need to count some pixels twice for that to work.

Let's make this simple (hope the formatting works):

Regular Bayer:
GR
BG

Sony's Bayer:
R
G G
B

Sony rotates the same patter 45 degrees counter-clockwise and then counts the GRG as three separate pixels as part of the same row, which is a bit like squeezing 3 people into a two seater, with one person sitting on the center column.

November 3, 2012

0
Reply
Thor

Formatting did not work - there was supposed to be spacing on the Sony pattern to indicate a diamond shape.

November 3, 2012

0
Reply
Thor

What reality are you talking about?

This is all about the specs used for TV and theatrical.

4k film full ap resolution is 4096x3112, but 440 pixels on the left is sound track. So it ends up 3656. I still hear people call it Academy 4k, even though it's not in reality 4k.

There are specs for some 4k TVs that are 3840x2160. Digital cinema projectors at 4k are 4096x2160.

With the digital cameras, the pixel counts are all over the place too.

These numbers are going to get only more convoluted as we add 8k.

Red once hinted on even higher than 8k resolutions as a possibility once. I don't know if we'll really want more than 8k, but then supposedly we don't need more than 4k.

November 5, 2012

0
Reply
Thomas

I'm talking about the reality where 4K means 4K - that is 4000. You do know what the K stands for, right?

4K is just the beginning. It doesn't match IMAX. 4K "works" for IMAX theaters, but it does not match the detail of a 15perf 70mm frame. It can't. So I think the proposed 9K sensor RED proposed in 2008 would be a good thing, assuming they ever make it. Hell, I'd like to see 12K and beyond. There are so many possibilities.

4K may be enough for regular theatrical presentations (2K really isn't), so if 4K distribution becomes the standard as hoped, then having the ability to punch in on 8, 9 or 12K shots without losing too much detail would be a nice option to have.

November 6, 2012

0
Reply
Thor

There is a lot of tech talk about this subject - pixels here and there, and so on. Who cares? The bottom line is that this camera is capable of stunning work unlike anything you may have ever seen before. Look at the many articles on the F65 and hear from those who are using it in the real world. There is nothing but supreme regards for this camera and how the data can be used to great advantage. That 16 bit 444 gives them an unheard of power and quality that is so rare. Colorists love it with passion.

Sony has built into it some future proofing by putting in a true 8K sensor. They are now stating that instead of downsampling to true 4k, they can make use of the full 8K in other ways possible. Sony is a huge company with a stellar reputation, enormous resources, and is dedicated to making some of the finest gear in the world. It is expensive - so what? You get what you pay for. Be grateful.

I'm pretty much betting that those of us reading these comments have to agree that it is a bit presumptuous to think that your opinions or so called knowledge supersedes that of Sony and their best technicians and developers in the world.

Cheers
Take 5

November 10, 2012

0
Reply

Hi Thomas, I calculated the 6k number based on the resolution listed in the F65 manual, 8192 x 2160. We can safely assume that number represents the total active pixels of the sensor. Sony would be crazy to list their horizontal pixels as only 2160 if there were actually more. Once we know the total active pixels equal 17,694,720 the resolution has be about 5782 x 3060 to fit in the 17:9 aspect ratio (also listed in the F65 manual.

http://www.talamas.com/Merchant2/EquipManuals/Sony%20F65%20Operation%20M...

January 6, 2013

0
Reply

Though it's definitely great that Sony is introducing new cameras (it is after all the move that has allowed me to afford an Epic :)) it seems to me like an overly complicated procedure which harbours artificial limitation at its core. Different recorders, with different media, with different acronyms all of which will likely carry exorbitant initial prices.
That said, it's really cool that they're releasing an open codec :)

November 2, 2012

0
Reply

Sony F65 SSD media is a lot cheaper (almost 50%) than same size RED SSD media... If you buy for example one terabyte cards.

Also, the new fast media for F5 / F55 will be cheaper than SRMemory used in F65.

So; unless RED really drops it's greedy media prices, Sony cameras, F5 / F55 will be a lot cheaper to shoot, when it comes to media.

November 2, 2012

0
Reply
Juhan-i

Actually this is not true and it has to do with the amount of data required.

This is where the compressed REDCODE creams the competition.

The difference is staggering. On a normal day it's not unusual to shoot 2-4TB (sometimes up to 8TB) of data on the F65, while on the RED Epic you'll usually stay under 1TB, even if you're shooting the industry standard 5:1 compression (although most TV productions stay around 8:1). At 5:1 you can fit over 30 minutes per 128GB card, whereas with the F65 you'll need the 512GB SRMemory card for 30 minutes.

So, while you can easily get away with using 4x 128GB cards per Epic, you'll need to use 4x 512GB cards per F65 - and then keep in mind you need to offload all this data and make at least two additional backups before reformatting the cards. You do not want a set full of cast and crew waiting for the memory cards.

So while the 1TB Sony SSD cards are cheaper per gig than the smaller RED SSD cards, you'll need a lot more of them, which takes longer AND requires even more storage space for archiving. Anything slower than a RAID tower will slow you down. USB2? Forget it., it would take you a full day just to back up one card.

There's a lot more to consider than just the raw cost per GB on the recording medium.

November 2, 2012

0
Reply
Thor

What no response on this one Juhan. *roll*

November 3, 2012

0
Reply
Lliam

I was actually more talking about the new F55 memory cards / 4K compression and how it compares to RED SSDs.

But, about the F65: you are wrong. Using the now available SQLite version of the RAW, you get about 160 MB/sec datarate for full 16-bit 24P 4K RAW material. There is no visual difference between original F65 RAW and the new RAW Lite.

So, one 512 GB card gets you about 54 minutes of material, not 30 minutes.

And that Sony card has full RAID5. So Sony actually gives users a more memory for same price and some of it is used to protect the userdata with RAID5 scheme. Something that RED does not do with its SSDs.

And btw, when I look at your example about RED storage requirements, it seems to be bit of too. One second of full 5K RED RAW (5120 x 2700 ) at 24 frames / sec, with 1:5 compression is about 127 MB / sec. After all, RED RAW is 16-bit dataformat.

That is, one 128 GB card stores 21 minutes of full 5K RAW material. Not 30 minutes. I chose this 5K resolution because it is closest what RED Epic has to offer, when compared to Sony F65. About 13,8 megapixels are used to form image. Sony on the other hand uses about 17,7 megapixels to form 4K image.

November 3, 2012

0
Reply
Juhan-i

Correction for above, actually it is even worse for full 5K RAW sensor data for RED. With 1:5 compression it is only about 17,3 miinutes for 128 GB card.

November 3, 2012

0
Reply
Juhan-i

Not sure what your beef is with RED, but it's clear your primary goal here is to spread misinformation.
So put your animosity aside for a moment and stick to facts, not opinions or spin. If there's one thing I hate, it's spin - and trolling, and you seem to be doing both.

Let's look at facts:
On a 128GB card, at 5:1 you can fit 36 minutes of 4K or 23 minutes of 5K. But we're talking about 4K here, which is why I said "over 30 minutes." This isn't a matter of opinion. Those are facts.

Also - the F65 numbers I used are for the 5Gbps compressed signal, not the uncompressed 19Gbps signal as it comes off the sensor (saving that signal is not an option.)
The lowest data rate for RAW 4K on the F65 is 1.5Gbps, which is 192MBps, and not 160MBps, just to be pedantic, at which you can record 50 minutes to a 512GB card, which is similar to a RED shooting 4K at 8:1.

BTW, SQLite is a database engine, so I have to assume you made an autocorrect typo.

As for how many pixels are being used to generate the final image, that's all well and good. If you want clean and fully measured 4K resolution, start with 5K or more. But don't say 17Mp equals 8K, because it doesn't.

The SRMemory cards are very good cards; I never said they weren't. But they obviously need to be good if you want to record 5Gbps without risking corrupt data.

The F5/55 looks like a pretty good setup, and it's about time Sony got serious about 4K for the rest of us, and I have to congratulate them on the global shutter (although skew hasn't really been an issue with the Epic in my experience) - but they're not out yet. Let's not compare available cameras to future cameras. Let's wait until they're out, then let's have a shoot-out and compare the two. Ultimately, the proof of the pudding is in the eating.

PS, those two new Sony cameras are competing with a now 2 year old RED product which features a 4 year old sensor, so assuming the F55 is out in time for NAB, let's put it up against the Dragon, shall we?

November 3, 2012

0
Reply
Thor

Nice footage but the point now is way to many acronyms, codecs, letters of the alphabet etc - what we need are SAMPLES from the F5 and F55. The F65 is still way beyond the reach and scope of 99.99% of NFS readers.

November 2, 2012

0
Reply
Paul

I have to agree with that, because after chasing every single video format and compression hardware over the last 20 years, technology is moving way too fast to justify the purchase of expensive cameras for independent projects. I went to film school, own DVX, 16mm and DSLR workhorses but, if there is a need, when the budget allows, I rather rent high end gear.

November 3, 2012

0
Reply
Agni Ortiz

8k? Please stop the madness.

November 2, 2012

0
Reply
Marcus

Having 8k is all well and good, but it doesn't matter if the image doesn't look good. And to me, especially from that video, I have yet to see something from the f65 that looks as good as the Alexa or the Red Epic even. It looks very video-ish.

I know that it hasn't been out for very long and there is only something like 100 been sold, but have any major features been shot on the f65 yet? It would be interesting to see if used properly could it look good.

November 2, 2012

0
Reply
Matthew Rogan

No trailers thought:

http://www.imdb.com/title/tt1815862/
http://www.imdb.com/title/tt1483013/

Here is one with the trailer:

http://www.imdb.com/title/tt1288558/

Difficult say about the image because the subject of the film. But I personally like it. Pro-work...

November 2, 2012

0
Reply
Juhan-i

In LA they are showing a 'music video' shot on the F65, projected in 4K before features. It looks, frankly, terrible. Shockingly bad.

November 2, 2012

0
Reply
marklondon

and I'm sure this is the camera's fault... mm hmm...

November 3, 2012

0
Reply
dixter

Well if I was Sony, I wouldn't be running it as an ad for the product. That's my point. They obviously saw it and approved it. and it looks terrible. Given i'm watching true 4k being projected 4k, on a film paid for by the camera company, I should have been blown away. Instead it was very VERY underwhelming.

November 3, 2012

0
Reply
marklondon

I have to admit, the video is pretty awful but, i don't think it has anything to do with the camera. Poor lighting, poor production design, uninspired framing/composition, maybe even poor grading but, hey, it's Taylor Swift. The reason Sony decided to use it is because it's her. And how would it sound if they didn't use it and later, it was leaked that it was shot on the F65?

November 3, 2012

0
Reply
dixter

and, not to mention that, after the dust settled on the Zacuto shootout, the F65 had the sweetest image after the Alexa.

November 3, 2012

0
Reply
dixter

I blown away by the demo footage on youtube. I don't care how many K's it is that looks really good.

November 2, 2012

0
Reply

That clip above looked pretty average to me. Her skin looks plastic. I loved the first footage they had up 18 months ago, but never seen anything good since. Have a gander at the Taylor Swift 4K clip i keep mentioning screening before features and let me know what you think.

November 2, 2012

0
Reply
marklondon

To me it was just a bit over average. I honestly feel like the image out of the BMCC made a larger impact on me.

November 3, 2012

0
Reply

I keep hearing about this Taylor Swift video but, I can't find it online. Anybody got a link?
Or. is it just in theaters?

November 3, 2012

0
Reply
dixter

Apparently you can also see it in-store: http://blog.sony.com/taylor-swift-sony-4k
I may go check that out at the Bev Center this weekend.

November 3, 2012

0
Reply
marklondon

@marklondon... if you let me know when you're going, I'll meet you there.

November 3, 2012

0
Reply
dixter

I don't think the skin tones look like plastic but I do think Taylor Swift's makeup is caked on pretty thick and she is very young so no flaws.

November 3, 2012

0
Reply

http://www.youtube.com/watch?v=s92smjLq_38

http://www.youtube.com/watch?v=y2jcgtsNC2M

http://vimeo.com/50643185

Last one includes the possibility to download slightly better version in 1080P.

November 3, 2012

0
Reply
Juhan-i

Yes, I've checked these out. That last one, shot by a serious DOP, is weirdly the least convincing. I don't know anyone in my professional life that would see that and say "oh, that's MUCH better - I MUST have it" than what they use currently. The DR is very nice, but in parts it looks very video. You shoot that exact film with an EPIC or Alexa, and if you're delivering 1080 the differences will be minor.
* full disclosure - I know the lighting company that provided the lights for that shoot.

November 3, 2012

0
Reply
marklondon

I agree with the "videoy" look. It might have something to do with the horrible grade they put on it. Color jumps from shot to shot and a very poor look over all. Also, the shadows looked quite noisy. I was not impressed with the camera or the work put into that film.

January 5, 2013

0
Reply

Quoting John Galt, Panavision Senior Vice President of Advanced Digital Imaging:
"Historically, 2K and 4K referred to the output of a line array scanner scanning film, so that for each frame scanned at 4K, you wind up with four thousand red pixels, four thousand green and four thousand blue..." ..."So if you have 4096 photo sites across the width of the film, in red and green and blue, and 3K along the height, you would have 4K by 3K. You'll have 12 million green photo-sites, 12 million blue photo-sites, 12 million red photo-sites. That's 36 million photo-sites. A 36 mega-pixel image is what you get from a 4K scan..." "you know very well that you cannot take a 8.3 million pixel sensor and create 36 million out of that without interpolation. You are up-converting, and there's really no value to the up-conversion. There's no new information."
"So 4K is not these 8 mega pixel or 9 mega pixel or 10 mega pixel CMOS images for the Bayer pattern where they add up all the pixels in a row and say hey, we got 4K. The great perpetrators of that mythology have been RED and Dalsa. That's why I call these "marketing pixels." It's intentional obfuscation. Because they really do nothing to improve image quality. They may improve sales volume. But they don't do anything to quality. But somehow the world has accepted that that's 4K. It's purely semantic. It's like saying, "I don't like my weight in pounds so I converted to kilos. It sounds better!" You'd be amazed at how many non-technical people I meet, often producers and directors, but sometimes even cinematographers get fooled by that stuff." Can anyone explain me if what Sony says had achieved proves him wrong or we are still interpolating what does not exist? Thanks

November 3, 2012

0
Reply
Agni Ortiz

Sony had to do something different, considering they've been railing against Bayer patterns for years. So, for the F35 (and Genesis) they used a striped pixel, whereby each of the 3 colors gets 33% of the horizontal resolution, which then gets interpolated based on the values of the other two to make up one "solid" pixel, except of course it never really worked and resulted in a lot of rainbowing artifacts. And they still poo-pooed the Bayer pattern.
So, when they made the F65, they again had to do something different. So they took a Bayer pattern and rotated it 45 degrees and used more than 4 pixels to interpolate the missing color information.

November 3, 2012

0
Reply
Thor

Those numbers of his are kinda shady...humans perceive detail based on the luminance of an image, not the chroma...we're really bad at seeing small changes in color, but good at seeing small changes in brightness. With clever math, you can use every pixel in a bayer pattern sensor to work out the luminance of an image, you just can't use every pixel for full color information. The difference between his "36 megapixel" image and an image from a 5k/6k/8k bayer pattern wouldn't be discernible once downscaled to 4k. He also ignores the fact that the film is still limiting the amount of information you actually get. You can use a trillion pixels to scan a piece of film in, but the film only contains so much data...and that's less than 4k.

To answer your question about Sony, the F65 only has ~17.5 megapixels (8192x2160 pixels), so it still falls short of his 36 megapixels.

November 3, 2012

0
Reply
Gabe

http://pro.sony.com/bbsc/video/videoChannelSearchResults.do/video-f65_te...

There is solid scientific reason why they allocate double phosites for green...

And yes, in "traditional sense" it is 6K Bayer-sensor. Allthought because green dominates the luminance equations, and even in human eyes (we can perceive more shade of greens than other colors), F65 has visibly higher resolution than for example RED Epic.

But you are not gonna see that by looking badly shot 4 Mbit/sec Taylor Swift Youtube-video.

November 3, 2012

0
Reply
Juhan-i

Ok, sorry. Link does not work. Try this:

http://pro.sony.com/bbsc/video/collections-technology_training/all-genre...

And choose the "F65 Technologies for 4K Cinematography"

November 3, 2012

0
Reply
Juhan-i

A standard bayer pattern has twice as many green photosites as red or blue. The F65 uses a bayer pattern rotated 45 degrees...any increase in resolution is from the number of total pixels, not the color pattern. Specs aside, the F65 has streaking and other visual defects in its footage...I'm they'll fix it in a firmware update, but as many Red haters like to say, resolution isn't everything. ;)

November 3, 2012

0
Reply
Gabe

When you say "visibly higher" - what does that mean exactly? Because to my eyes, a RAW 5K Epic frame is higher resolution than a RAW 4K F65 frame so I'm curious as to what you mean by "visibly higher" - without getting into a resolution war.

BTW, the 4K F65 RAW isn't truly "RAW" in the strictest sense of the word because it has been processed (sub-sampled down from 6K.)

November 3, 2012

0
Reply
Thor

Ok, this is interesting discussion, but I can not answer to everyone, do not have time. I put my all my answers here.

Gabe : did you watch the presentation ? There the Sony scientist spends rather long time trying to describe their approach. Main point is that F65 sensor has full resolution (4K) of green photosites.

Every single green component in pixel in final visible 4K image matrix has unique phosite. So green information actually is not interpolated. Again, both blue and red have half the resolution. So they are interpolated like with ordinary Bayer-sensor arragement. So, the F65 sensor is more sensitive to green color, just like human eyes. Because there is more resolution when sensor measures the green color intensity values from outside world.

And when it comes to problems with F65 sensor, well I have not heard anything about it...

It sounds bit strange considering all the people who actually have worked with F65 usually say that it is THE digital camera at the moment. Of course you can always read reduser.net for different expert opinions. Like your claim about problems with sensor... Or maybe it was you who actually wrote it there.

So, maybe I should put quote here too from reduser.net :

http://www.reduser.net/forum/showthread.php?86185-People-not-getting-the...

There you go. Of course after this comment from real professional, not single red-boy had anything say anymore about F65.

Yes, so called "peer pressure" is a terrible thing: even when you know truth about something, you have to downplay it so that others do not think that you are...Sony pussy... :

http://www.reduser.net/forum/showthread.php?87222-Worthy-competitor-for-...

And about the user "Thor" asking the resolution, well F65 17+ million light sensitive photosites that cover S35 sensor area. Red Epic has about 13,8 million light sensitive photosites that cover slightly larger sensor area. How can Red Epic have more resolution if you put same lense on both cameras ?

It is not the final image resolution that is important, it is the amount of data sensor covers, senses from the incoming light. When you view that 5K image from RED, you still have to downsample it to 4K.

If you don't believe me (which of course is fine), you could try find the french discussion about F65 / Epic. It was thread in french cinema board where they had those standard test charts shot with both cameras. Maybe there was other cameras too, don't remenber.

But real world results were clear, F65 has a lot more resolution than Epic. Maybe I could find that thread again.

November 3, 2012

0
Reply
Juhan-i

Interesting - you didn't actually answer my question, you just restated your previous assertion.

I asked you to clarify what you meant by "visibly higher" - which has nothing to do with the number of photosites on the sensor.

Also - as a filmmaker, I can unequivocally tell you that it IS about the final image and nothing else. Why do you think so many DPs choose Alexa when it has a lower resolution than either Epic or F65? It's because they prefer the final image it produces. The F65 is not THE digital camera out there. I have not heard anyone say that of any of the people I have spoken to who have used it. They like it, but most of them still prefer either the Alexa or the Epic.

You don't HAVE to downsample the Epic's 5K image, you can just stay in 5K and it looks great. Why? Because it has luminance information for the full 5K frame, and only subsamples the chroma. Measured resolution is around 4K, maybe a hair over - but of course, if you downsample to 4K you'll have to apply anti-aliasing filters which will reduce the measured resolution. A 4K image will never measure a full 4K - not ever - unless you're willing to live with jaggies in sharp lines - at which point it no longer looks like film and we've defeated the purpose.

Also - the argument that every single pixel in an F65 file has a corresponding green photosite is misleading at best, outright false at worst. Each pixel is made up of more than one green photosite and the adjacent red and blue photosites, using interpolation to fill in the gaps. That's exactly what a Bayer pattern does. The F65 sensor is no different, even if it has more green photosites in total - because it has more total photosites.

And finally, if HALF the photosites are either blue or red with the other half being green, how on earth can you claim the green channel is not interpolated? It would be full of holes if it wasn't. For the love of everything good and pure, stop the fucking spin. You're being a Sony fanboy - and it's irritating.

Whether or not someone believes you is irrelevant - facts are relevant, and you're ignoring them in favor of Sony's spin and hyperbole machine.

PS, Thor is my real name.

November 3, 2012

0
Reply
Thor

I'm afraid you're still buying into marketing hype...all bayer sensors are more sensitive to green like human eyes, because they all have double the green photosites. All the F65 does is rotate the bayer pattern 45 degrees, and from what I understand this is to get a better image when using less sophisticated interpolation methods. There are pros and cons to that, but just like the F35's solution, there's no free lunch.

The F65 is certainly a beast, but at the end of the day no one's picking it over the Alexa and Epic. That's a big part of what Sony's trying to fix with the F5 and F55.

November 3, 2012

0
Reply
Gabe

Why would you say the green channel is bein interpolated?

If its a 8k sensor with 4k of green and outputting 4k, where is the interpolation?

Juhani is a Sony fanboy and you are a Red fanboy. Fair enough. It's spin on both sides.

November 3, 2012

0
Reply
Thomas

Thomas - because the green photosites on the sensor are not in the same physical location as the blue or the red, which means if you want to have green information for those pixels too, you need to interpolate.
It's not the total number on the sensor that matters, it's the ratio between green and red and blue. There are twice as many green as there are of either red or blue - so obviously you need to interpolate all 3 colors, just less with the green. If you don't interpolate the green, then the green channel will literally have gaps in the information, even when scaled down.

As for being a RED fanboy - I may come off that way because I'm trying to refute the anti-RED FUD, but believe me, I'm just interested in facts, not spin. RED has been guilty of spin too - but they also have Graeme Nattress on their team and he doesn't deal in spin. He deals in math.

November 3, 2012

0
Reply
Thor

Pages