Shootout: RedOne vs F35

[danbrazda]

I'd love to see a side-by-side shootout between these two cameras. Same lighting, lenses...etc. Would take a lot of connections to get the 2 cams in the same place at the same time but as diverse as this forum is maybe someone can pull it off. Just curious if the differences would be apparent enough to justify the cost of the F35. Any conjectures?

[Steve Phillipps]

One thing that concerns me about the Red is the compression, ie 36 and 28 mb/sec. I confess straight off that I'm by no mens an expert here. I know wavelet codecs are supposed to be good, but it seems there's a big drop from the 130 mb/sec or whatever HDCam is, to 36. And if you're downloading from 4k sensor instead of 2/3" surely that means even more compression to fit into the bandwidth? Having said all that, by all accounts the images look stunning, so great! There have been reports of graininess here and there and a few other little issues that could be result of high compression?
Feel free to correct me if I'm entirely wrong, I won't take it personally!
Steve

[Sidney L. Plaut]

I talked to a rental house here in copenhagen - www.ebh.dk. They had seen
f35, HPX3000, regular 35mm and red1 footage, all "video" scanned to 35mm.
I think maybe the test was done in Norway... maybe ask some of the norwegians in the forum if they were involved with it...

The rental house - and this is ofcourse completely "hearsay" - that the RED1 blew all the videocams away. 35mm is another matter. Film is film... that comes down to an aestetic choice - and ofcourse financially:)

[esmilis]

One thing that concerns me about the Red is the compression, ie 36 and 28 mb/sec. I confess straight off that I'm by no mens an expert here. I know wavelet codecs are supposed to be good, but it seems there's a big drop from the 130 mb/sec or whatever HDCam is, to 36. And if you're downloading from 4k sensor instead of 2/3" surely that means even more compression to fit into the bandwidth? Having said all that, by all accounts the images look stunning, so great! There have been reports of graininess here and there and a few other little issues that could be result of high compression?
Feel free to correct me if I'm entirely wrong, I won't take it personally!
Steve

I don't know how much HDCam is but red is 36,28MB/s (288, 224mbps). And thats a lot

[Steve Phillipps]

So am I getting my bits and bytes mixed up here? HDCam is 144 megabits per second and Red is 36 megabytes per second (ie 288 megabits)?
I thought I was an expert until 3 minutes ago!
Steve:waaa:

[Stephen Williams]

Hi,

Claudia Miranda tested Red v The Sony F23. Red did not blow the F23 out of the water, some people were very offended by the results of a real world test.

Stephen

[Richard Lackey]

The F35 has a 35mm sized sensor right?

[Steve Phillipps]

Was the Red expected to blow the megabucks Sony out of the water?

[Stuart English]

So am I getting my bits and bytes mixed up here? HDCam is 144 megabits per second and Red is 36 megabytes per second (ie 288 megabits)? I thought I was an expert until 3 minutes ago!

Yeah, but its O.K.. we are all learning :construction: and something else that is non-intuitive - as the resolution goes up the ability to compress gets easier. So there are a whole bunch of reasons why you can't simply extrapolate what you know about existing cameras and compression and apply that to RED.

[Mark Tyson]

Folks;

It would be folly to assume Sony can't make a camera for 200K that
might be able to compete...maybe even favorably...with RED. No one
should mis-underestimate (as our political leadership might say) the
engineering abilities of other camera manufacturers. Just revel in the
cost/quality convergence that RED offers. If you have the bucks to use the Sony F35, your concerns are much different than most of us.

Mark Tyson

[Júlio Taubkin]

Hi,

Claudia Miranda tested Red v The Sony F23. Red did not blow the F23 out of the water, some people were very offended by the results of a real world test.

Stephen

Hi Stephen, you spelled Claudio's name wrong there...

Cheers! :)

[Vincent Rice]

I imagine the pictures from the F35 will be stunning, with a wide dynamic range. I understand its a full 35mm 12Mp CCD sensor being downsampled to 4:4:4 1080p.

Of course the F35 will probably cost $300,000

[Bing Bailey]

its the sony version of panavision genesis camera and it shoots great images but it looks really sterile. there is such a thing as too clean. it feels like all the texture or the feel of texture is gone. it makes people look a little plastic. at least it did in superman returns. red images feel more organic , they have texture, not as much as film but it's early days yet

[chuckt]

The F35 has a 35mm sized sensor right?

"F23 Chassis with 35mm full aperture size striped CCD sensor and PL mount, camera has 1920 X 1080 output. Currently available only in Japan from Omnibus, rest of world after late 2008/Early 2009. 50P 4:4:4 maximum frame rate vs 60P 4:4:4 with F23. F35 is 1 Chip with Stripped CCD array vs 3 dedicated RGB sensors on F23. 35mm DOF vs 2/3" with DigiPrimes DOF."

[Lappe767]

http://www.solarsystemstudio.com/SolarSystem/Blog/Entries/2007/12/9_Red_Digital_Cinema%E2%80%99s_Red_One_vs._Sony%E2 %80%99s_CineAlta_F35.html

[Greg M]

Hi,

Claudia Miranda tested Red v The Sony F23. Red did not blow the F23 out of the water, some people were very offended by the results of a real world test.

Stephen

why would they be offended, its an apple to oranges comparison. The F23 is 10X the price of a Red One, so anyone expecting Red to "blow it out of the water" was dreaming. Reality is they are two different animals, Red One is 4k, F23 is 1080....the F23 is tape based vs data based. I have used the F23 and it is a fine camera, the dynamic range is incredible, and its current workflow works. Once Red delivers the SDX and opens up the files to other post solutions then Red will have the post advantage, and at $18k it is one hell of a camera that can compete nicely with the very expensive Sony's.
In a few months I suspect the Red One will close the gap on quality issues.

[Clint Johnson]

So am I getting my bits and bytes mixed up here? HDCam is 144 megabits per second and Red is 36 megabytes per second (ie 288 megabits)?
I thought I was an expert until 3 minutes ago!
Steve:waaa:

It's all in the capitalization and most folk tend to mix and match to ill effect. MBps is megabytes per second and Mbps is megabits per second.

[Craig W. Bickerstaff]

I think it would be interesting to do a short film or whatever only you shoot with various cameras and intercut them, Say some of it on the f35 some on RED ONE other stuff on D21. Maybe even a few shots on 35 mm film.
Use all the same glass on all of then have people try and guess which shots came from which camera. I think if you shot if right it would be too seamless to tell.

[laguun]

Almost all HDCAM Cameras can record uncompressed via HD-SDI.

The 750/790/900 (etc) record uncompressed YUV422@10bit, the 950/F23 (etc) and probably F35 RGB444 as well.

To avoid the HDCAM 144-180 Mbit (depending on framerate) or 440-880 HDCAM SR is recommendable for VFX / key etc, but not necessary. The same applies to redcode.

However colorists and VFX artists often require more time for the same work, and can get less excellent results, if massive datareduction is in the source.

The resolution bottleneck, if distributed via film, is 35mm projection and the interpositive/negatives anyhow in both cases.

We, being HDCAM owners and users since 2002 prefer RED over HDCAM incl. F23 for many scenarios and vice versa.

The Imagequality in resolution is much better with the red, and having a raw workflow is excellent. The flexible framerates are excellent for many creative scenarios, and having 35mm DOF and PL-mounts is very good. The formfactor and the size of red is much better than F23/genesis/F35. Having recordable

However, the still constrained red workflow is rather unflexible and slow for most postproduction systems and houses who different than us dont have a specialised red-workflow. Until the support of these systems will be allowed, while HDCAM is widely supported, completely stable and unbuggy and realtime throughout, this is therefore a 50/50 neutral / depends on the situation thing for us.

HDCAM also had advantages, as top-quality 1080p live, exchange media (tape), a slightly higher light sensitivity especially at interlace, and for some scenarios the 2/3 is in fact an advantage (one-man crew animal, long zoom etc), also hdcams CCD-FITs cameras are less depending on light temperature and have no skew.

Therefore we are very happy to use both systems, and i often prefer red when i can control light and can go wild on the creative side (overcranking etc), while HDCAM offers benefits for international doc, live, fast turnarounds, exchange etc.

I am very happy that red offered a excellent alternative to hdcam, however we dont feel a reduction of the value of our hdcam - in fact the class 1 crts, hdcam (sr) VTRs etc are often plannend and used additionally also when using red. And hey, its BIG fun to have a red or a hdcam as making-of camera when shooting in hdcam or red :)

[Bang WOW Bang]

Yeah, but its O.K.. we are all learning :construction: and something else that is non-intuitive - as the resolution goes up the ability to compress gets easier. So there are a whole bunch of reasons why you can't simply extrapolate what you know about existing cameras and compression and apply that to RED.

many DTV broadcaster used HDCAM 108oi only as their HD master then recompressed to either MPEG-2 or H.264 in their head-end down to 13mb/s and the 1080i pcitures look Crapped !!!

Even in HKG like this. Double Compression is no good unless the 1st compression is RAW material. RECODE capture to RECODE display is the way of 4/3/2K to some serious watchers then average TV watchers.

Stuart, did you listen to SRV music in TX, and I am a big fan of him and Freddie King.

Cheers mate.

Stewart
HKG

[Tom Lowe]

The rental house - and this is ofcourse completely "hearsay" - that the RED1 blew all the videocams away. 35mm is another matter. Film is film... that comes down to an aestetic choice - and ofcourse financially:)

I seriously doubt that any HD camera will blow the F35 out of the water. It's a very good camera and very expensive.

[David Wyatt]

I'd wait for Build 16 before any meaningful image comparisons are made...

The F35 is just the Genesis by a different name - not particularly stunning considering it'll set you back close to $300,000 (I'll have 17 Reds instead please :biggrin:)

[Craig W. Bickerstaff]

I'm not sure that the F35 is just a Sony branded Genesis I think there is more to the camera than that.
And I think that in the FreshDV interview the price is still undecided but will probably be around $200,000 to $250,000

[laguun]

I'm not sure that the F35 is just a Sony branded Genesis I think there is more to the camera than that.
And I think that in the FreshDV interview the price is still undecided but will probably be around $200,000 to $250,000

I am also not sure its a genesis, however the sensor specs match exactly.

[Mark Andersen]

In really simple terms the plastic look is in the "video" genes. The Red is more like a digital still camera that shoots really fast, than a video camera. This is why it feels more organic. The RAW system of compression is very different than video compression so it is not fair to compare bit rates of recording. RAW is much more efficient but at a price, that price is rendering time (pixel interpolation) still photos shot raw go through the same process. In digital stills shooting RAW is the highest quality you can get.

[sparkhope]

I am also not sure its a genesis, however the sensor specs match exactly.

The engineer at NAB said it was an updated Genesis for whatever that's worth. One thing that got me however, is the viewfinder...it's a removeable mirror mechanism that reflects an image from a small lcd (approx. 4 inches wide). The kicker is that the lcd doesn't give you an hd image, not even 720.

For the money I really expected innovation on that front.

[Emmanuel Cambier]

I'd wait for Build 16 before any meaningful image comparisons are made...

The F35 is just the Genesis by a different name - not particularly stunning considering it'll set you back close to $300,000 (I'll have 17 Reds instead please :biggrin:)

Build 16 is Jim's answer to this very test, I believe.:ninja:

Emmanuel

[Bang WOW Bang]

I am also not sure its a genesis, however the sensor specs match exactly.

is it worth just recorded on HDCAM SR format COMPRESSED or we have options like S.two or Codex uncompressed RGB444.

David Fincher did that uncompressed HD444 gig but with Viper Cams ( 6 units for one project )

Stewart
HKG
WOW Holdings

[Pierce Cook]

http://www.solarsystemstudio.com/SolarSystem/Blog/Entries/2007/12/9_Red_Digital_Cinema%E2%80%99s_Red_One_vs._Sony%E2 %80%99s_CineAlta_F35.html

The first paragraph at this link reads as follows:

"Sony is trying to block RED from being used on Sony Pictures movies. Panavision has also made it their policy not to rent lenses to Red One owners. I could have made this a Red One vs. Panavision Genesis comparison, but since the Genesis is only available for rental, the price list and availability wouldn't be relevant."

What's up. I know there has been some backlash, to say the least, but that seems unprofessional (life's not fair, I know).

What's the deal.

[Matthew Verkler]

I checked out the F35 and the F23 at NAB as thoroughly as I could. The main differences between them and the Red as I see them:

-Cost. Obviously.

-Form factor. The Sony's are big and heavy. For some that won't matter, but it's important to me.

-Nature of the image. It's been said before here, that the Sony's take very clean pictures. True. It's also been said that they look antiseptic. Also true, IMO. The Sony's have been referred to as having video genes, and that accounts for their look. I somewhat disagree, as the Panasonics all have a very pleasing look and colorimetry compared to the Sony's, from the Varicam and DVX to the HVX and the HPX's. And Red is much more so, and as it has been said, organic. It has always seemed to me that Sony has designed their imagers to look good to test instruments, while others, like Panasonic since 2000, have gone for what looks good to the human eye, at the expense of some "cleanness". It's all subjective of course, and we choose the tools that support the story we are trying to tell. Red fits my aesthetic very well, though I can conceive of a situation where I would favor the Sony's - perhaps some kind of THX 1138 kind of film or the first part of "Joe Versus The Volcano."

I like the fact that there are choices available to us; for me, I chose Red, and have been ecstatic with what I have been getting out of it. And the workflow isn't turning out to be a problem either. Shooting P2 also required a lot of changes to how we do things, and now that I am used to that, I really don't ever want to shoot on tape again. P2 has been like the "gateway drug" to harder drugs like Red!

On another note, the more I think about the coming of Epic, the more I am stoked. When I first heard about it, I thought, well I certainly don't need that, as I am very satisfied with my Red One and don't see the need for an upgrade. But after seeing the form factor and analyzing the proposed specs, it's lighter, smaller, has more recording options for bitrate, newer sensor. AND, I can turn in my Red One body for full credit ($17500) if I choose to. THERE IS NO DOWNSIDE TO THIS! Yes, it's more expensive, so I'll cross that bridge when I get to it. Perhaps I won't want to spend the extra money. But seriously, the money I spent on Red One will all come back to me if I get the Epic? That's like a free rental of the Red One body. Unbelievable.

And anyone who complained about Epic, think about it: you should be ashamed of yourselves. There is NO downside!

[Sidney L. Plaut]

Tom,

You got a point.. I need to test it myself - and not quote "rumors."

The f35 is prob. a great HD Camera:)

and I admit Im abit biased towards the RED:)

[Dylan Reeve]

The first paragraph at this link reads as follows:

"Sony is trying to block RED from being used on Sony Pictures movies. Panavision has also made it their policy not to rent lenses to Red One owners. I could have made this a Red One vs. Panavision Genesis comparison, but since the Genesis is only available for rental, the price list and availability wouldn't be relevant."

What's up. I know there has been some backlash, to say the least, but that seems unprofessional (life's not fair, I know).

What's the deal.

Sony Pictures are not keen to shoot a major film on a camera that is still essentially in beta test, quite reasonably. Once the firmware builds are stabilised a little, and the post workflow is a little more solid then I doubt any such restriction will be enforced.

As for Panavision, apparently that's be debunked also, there've been a number of posts here stating that it's definitely not the case. Some PV branches may want to limit their rentals, as they don't want to be left with cameras but no glass to rent for them, so they'll probably limit the rental of lenses if it will leave them with insufficient inventory to support their own camera stock.

[Jeff Brue]

We did David Fincher with Claudio Miranda F23/Red test. We have both cameras, they both offer different imaging capabilities. Guys they're film stocks at this point, they have different looks. The daylight footage of that test did amazingly well, it was the night stuff that looked bad (I debayered it prepped it and was in on the session). Red has since made massive strides, and I have not received any night footage since that has been on that level.

F23 still has more range in the highlights, but well come on, don't go bashing or making base judgements on tests you've heard about. All things being said the only reason we didn't release that footage was because in our opinion Red changed enough as far as sensor capabilities that releasing the footage was irrelevent.

Now the Build 16, F23, and D21 test we're planning that I want to release.

[Pierce Cook]

Sony Pictures are not keen to shoot a major film on a camera that is still essentially in beta test, quite reasonably. Once the firmware builds are stabilised a little, and the post workflow is a little more solid then I doubt any such restriction will be enforced.

As for Panavision, apparently that's be debunked also, there've been a number of posts here stating that it's definitely not the case. Some PV branches may want to limit their rentals, as they don't want to be left with cameras but no glass to rent for them, so they'll probably limit the rental of lenses if it will leave them with insufficient inventory to support their own camera stock.

That I completely understand. Perhaps I took the original statement more pessimistically than I should have.

[Dylan Reeve]

That I completely understand. Perhaps I took the original statement more pessimistically than I should have.

I think a lot of people did. People seem to be looking for conspiracies.

[Graeme Nattress]

I'd have loved to have pointed the F35 at a zone plate to see how it's true resolution measures, and how much aliasing gets through. That's the real benefit of doing a 4k Bayer pattern sensor - you can properly optically filter and still have great resolution.

Graeme

[combatentropy]

Yes, I think Bayer is clever.

A Bayer sensor normally keeps half of its total pixels when converted to full color (a 12 megapixel Bayer sensor winds up with roughly a 6 megapixel RGB final image). Meanwhile a sensor striped in columns of red, green, blue, red, green, blue -- like on the F35 -- keeps only a third. Bayer can be thought of as a form of "compression" before you compress digitally.

Furthermore, I think wavelet compression is stunning. Some people have reported testing a compression of 40 to one without being able to see any difference from the original. I've eyeballed full-resolution stills from the Red. No jaggies or mosquito noise like you may have, say, in a JPEG. If you have a hard time detecting artifacts while slowly examining a still image, how much less noticeable will any be in the rapidly moving images of an engrossing story?

[Craig W. Bickerstaff]

I think I prefer the striped approach as you get full color reproduction as to my understanding.

[GlennChan]

I don't think the striped approach gives you "full" color reproduction since the elements are still in different locations / offset from each other. Ideally you'd want the elements stacked into the same location *with perfect optics and spectral response*.

Some lenses will have some chromatic aberration, so it is one barrier to full color reproduction.
I believe the 3CCD prism will introduce some CA. (Though many 3CCD cameras use pixel shifting, which is probably not the "full" color reproduction you're looking for.)

Foveon doesn't have the prism problem, but then it has its own problems since silicon is a poor color filter.

2- In practice I don't think all that stuff matters too much. Human vision doesn't have as good color acuity as luminance acuity... so it's not really an issue if luminance resolution is better than color resolution.

The Red makes good images.

And it even chroma keys really well. (Better than some other 3CCD systems out there.)

So I don't think there are really much issues with the color reproduction (in the spatial/resolution /artifacts sense).

[sander kamp]

We did David Fincher with Claudio Miranda F23/Red test. We have both cameras, they both offer different imaging capabilities. Guys they're film stocks at this point, they have different looks. The daylight footage of that test did amazingly well, it was the night stuff that looked bad (I debayered it prepped it and was in on the session). Red has since made massive strides, and I have not received any night footage since that has been on that level.

F23 still has more range in the highlights, but well come on, don't go bashing or making base judgements on tests you've heard about. All things being said the only reason we didn't release that footage was because in our opinion Red changed enough as far as sensor capabilities that releasing the footage was irrelevent.

Now the Build 16, F23, and D21 test we're planning that I want to release.

Very sensible comments from someone who knows what he is talking about. Thanks a lot and looking forward to the new test!

[Jimmy Shen]

I watched both Red and F23/F35 footage projected 4K at NAB a few days ago.
I wasn't that impressed with the F23/F35 stuff they've shown. Some shots looked very video-y. They intercut the F23 with F35, I couldn't tell which is which.
From what I've seen, I much preferred the Red's image quality.

[Paul Leeming]

I'd love to see a side-by-side shootout between these two cameras. Same lighting, lenses...etc. Would take a lot of connections to get the 2 cams in the same place at the same time but as diverse as this forum is maybe someone can pull it off. Just curious if the differences would be apparent enough to justify the cost of the F35. Any conjectures?
It's been done, with Red #378 against one of Omnibus' F35s. Unfortunately we weren't given the F35's results so we can't definitively say one was better than the other. I can tell you Omnibus were impressed by Red and probably had a slight case of buyer's remorse after spending that amount of money on three of the F35 beasts!!

If anyone particularly wants this comparison to be done publicly and is willing to pay for the F35 rental we'd be happy to organise it.

Paul

[brandon herman]

Red fits my aesthetic very well, though I can conceive of a situation where I would favor the Sony's - perhaps some kind of THX 1138 kind of film or the first part of "Joe Versus The Volcano."

You don't see those two films mentioned in the same sentence very often.

But it made me smile.

:biggrin:

[Graeme Nattress]

Sure, an RGB stripe gives equal resolution to each of RGB, whereas a Bayer pattern gives higher luma resolution at the "expense" of chroma being a lower resolution. However.....

On a bayer pattern we know the colours are not co-sited. One of the tasks of the demosaic is to make the colours appear co-sited and lined up nicely on top of each other. That's one of the reasons the demosaic can reduce the absolute resolution a little bit. However, it's a known issue and we deal with it. In an RGB stripe, you sort of just assume the colours are co-sited when they're not, just like running a pixel shift 3 CCD on 4:4:4 you assume co-sited when they're not.

The only real issue with a bayer pattern is chroma moire. We've seen it on very sharp test patterns, but we also have our chroma de-moire tool too, so it's hardly a practical issue for "normal" shooting subjects.

Camera design is the art of compromise. To me, the Bayer pattern is still the best compromise, when you use it with a full understanding of how it works.

Graeme

[David Wyatt]

I'm not sure that the F35 is just a Sony branded Genesis I think there is more to the camera than that.
And I think that in the FreshDV interview the price is still undecided but will probably be around $200,000 to $250,000

Is that including the deck? OK, I'll have 14 Reds instead :biggrin:

[michael zaletel]

I'm just gonna say this is one of the coolest threads I've read here in a while. Tons of smart, experienced, passionate people hashing it out.

Will F35 owners/users ever have such a forum? I think not.

-shooter

[Stephen Williams]

why would they be offended,

Hi Greg,

No idea but there was a great deal of noise.

Stephen

[rod bradley]

Very smart people indeed, hashing out which dog is top dog/better dog. And I am happy for their passion and that they have the time to do it.

Many of us who have Reds (if we're fortunate) are working to pay for a camera that we frankly don't care much whether it's better than such and such, we know it's an astonishing tool and allows us to take risks using cinema expression on a higher quality plane than we could've afforded a year ago. For that I am grateful, whether it's better or not better or just different from the F23 or F35 or the D20 or D21. And maybe they are all great innovations and remarkable tools that are all plenty fine enough to do the job. There are Canon fans, Nikon fans, Pentax fans, Olympus fans -- I've used them all, and they all worked and when I look at the photographs, I don't really care or remember which tool I used.

But of course, let's keep testing and pushing the geniuses behind the technology to continue astonishing us. Then put our astonishment to good use making this very tiny part of human expression work toward making the world a bit beter place.

[RED,BLUE&GREEN]

This is a great thread. Very intelligent notes and comments.

A couple of things after going to NAB and talking to head Sony engineers,Ted from Red, Band pro, and a lot (and I do mean a lot) of DPs and shooters.

The Sony F23 is now at about 120k for a body and viewfinder...not 10 times a red, but still a lot. ( remember with red you need handles,,,etc..etc...not just 17,500)

The Sony Viewfinder on both the F23/35 is at least 720 . And the cameras have a true (HD) output 1080p.

As far as tests being done. A Dp from NY (Du art) shot Redone and Sony ex-1 footage and projected them both at 4k at the Sony booth ( 1pm on tuesday april 16th ) Some liked the EX-1 better ( maybe because we were at the sony booth)..others like myself, thought they were very close. Thats a 7,000 camera. About 1/3 the cost of the Red.

The F35 Sony Does have a Genesis imager. But as far a technology... thats about it for being the same. 5 years of development have gone into this camera. Its like saying build 1 to build 16 plus a few years! ( fyi Sony is building an EXMOR type chip for the F35, 2k and next years 4k systems. It might be ready when the f35 is released)

I will be shooting a test ( that is fair!!) when build 16 and the f35 are both ready. This will be done with arri Master Prime ( full frame cameras) and Digi Prime (for 2/3's cameras) and of course we will presenting this first without titles of which camera is which.

I guess because technology is so great these days HD or 4k or whatever is in the eye of the beholder. I think the F35 footage is the best cleanest 2k or for that matter 4k footage I have ever seen. But clean or noiseless may not be the best to everyone. I cant wait. I have been waiting for a great 35mm 4k camera for a long time ...build 16...bring it on!!

And for the Future...the best picture I saw at NAB was the 8k camera prototype by NHK and Sony....talk about resolution!!! 34mega pixel camera!

[RED,BLUE&GREEN]

By the way RodBradley.....

Spoken eloquently! Here here!

[Dj Joofa]

Furthermore, I think wavelet compression is stunning.

Combatentropy, I apologize in advance if the following message is terminology heavy but I wanted to provide a clearer picture why something more important is at play here and is not mentioned frequently, which Wavelet compression has taken advantage of very nicely.

Most Red users will be filming natural images. The power spectrum of natural images falls of more like 1/f^2, where f is the frequency, assuming that image statistics are at least wide sense stationary. For natural images, for which as mentioned, certain important image statistics fall off as square of image frequency, it will be helpful if we can capture this fall behavior using some ideally linear transformations. And, transforms such as Karhounen-Loeve / PCAs /etc., do capture this information very well, which is in accordance with certain Markov processes.

Wavelet processing is closely related with band-pass filtering, which electrical engineers have been doing for ages, and hence, such band-pass behavior ties in nicely with capturing the image power fall off, though, perhaps, not as ideally as KLT / PCA. However, it is reasonably good.

Therefore it is the properties/statistics of natural images that Wavelets is able to exploit more usefully than the often-used DCT/DFTs that are helping us here. On the other hand it is possible to provide example of images where Wavelets performs poorly because they could not exploit certain image statistics.

[Matthew Verkler]

You don't see those two films mentioned in the same sentence very often.

But it made me smile.

:biggrin:

Very glad to be of service. Smiling = good.

The first part of "Joe Versus the Volcano" really stuck in my memory; the fluorescent sickly image and the ballast buzz sound to go with it. I'm so glad I don't work in a regular office!

[Peter Moretti]

...
As far as tests being done. A Dp from NY (Du art) shot Redone and Sony ex-1 footage and projected them both at 4k at the Sony booth ( 1pm on tuesday april 16th ) Some liked the EX-1 better ( maybe because we were at the sony booth)..others like myself, thought they were very close. Thats a 7,000 camera. About 1/3 the cost of the Red.
...This is where test get very hard to evaluate unless you know exactly how they were done. The EX1 can spit out uncompressed, 10-bit color out of its HD-SDI port. I bet that looks amazing. So how were both images recorded? And what was the subject matter? Were there a lot of highlights? What lens was used on the Red? What stop on each? How was the DoF? I could go on and on as you obviously know. I have no doubt that under certain conditions an EX1 can make nicer images than a Red One... and of course vice versa.

[Zk2007]

also hdcams CCD-FITs cameras are less depending on light temperature.


Could you please elaborate on that? Is this a case of CMOS vs CCD or a bayer problem?

[CK Olsen]

So am I getting my bits and bytes mixed up here? HDCam is 144 megabits per second and Red is 36 megabytes per second (ie 288 megabits)?
I thought I was an expert until 3 minutes ago!
Steve:waaa:

Back to Steve's point, for a minute...
When talking about specs, remember the pixels:

HDCAM is 144 Mbit for a 1920x1080 final pixel image...
RED is 288 Mbit for a (roughly) 4096x2048 final pixel image...

On the surface, the RED frame appears to have 4x more pixels, but only 2x the data rate.

It's certainly not an apple to apple comparison, but it is important to remember that despite how pretty the frames look, there is a substantial amount of data gettting squeezed down into the final Redcode 28 or 36 stream.

(erm... but please review Graeme's post below for an accurate breakdown of the data... Graeme's Da Man!)

[Graeme Nattress]

HDCAM is storing 1440x1080 luma (=1555200) + 2 * 640 * 1080 (=1382400) chroma "pixels". That's a total of: 2937600 8bit "pixels", or 23,500,800bits.

We store 4096x2304x12 bits = 113,246,208bits, or nearly 5 (4.81) times as much.

However.... And this is the key part....

Wavelet based codecs are more efficient than DCT based codecs. RAW is easier to compress than "video", and wavelets get more efficient on larger frame sizes! Put that all together and that's why REDCODE RAW 4k images look much better than HDCAM on only around twice the data rate. That and the magic pixies.....

Graeme

[CK Olsen]

Now talk about helpful!
Glad to see you on the boards Graeme, and thanks for clearing all that up!

[Tom Lowe]

I think proportionally, I learn more from Graeme's posts than any other reduser, even if I don't understand half of it. :)

[Graeme Nattress]

Thanks Tom, I think.... But this is a forum, so if you don't understand, I'll try and help with the missing info to make it all make sense.

Graeme

[Tom Lowe]

Well that's no reflection on you, Graeme! I just don't have a good knowledge base with a lot of this stuff to begin with, so often when I read your posts I end up on wikipedia or whatever trying to get up to speed on the basics of what you are saying. :)

Mullen is kind of the same way. Even after two years of subscribing to American Cinematographer, I only undertsand about half of what he says, just because it's so detailed and requires some base of institutional knowledge.

[Christoffer Glans]

Take into account that RED got the technology blueprints from 20 000 year old cave drawings from a civilization that was originally run by ailens with superior technology...

I mean, other companies just can't compete with that, simple as that.
Also why they don't talk about the mysterious sensor.

Hope I don't get in trouble for revealing the truth? :P

[Dj Joofa]

RAW is easier to compress than "video"


But isn't RAW linear light and and video gamma corrected? Hence, video is perceptually more uniform and should be able to compress better.

[Graeme Nattress]

Joofa, that's where the magic pixies come in!

Graeme

[Dj Joofa]

Joofa, that's where the magic pixies come in!

Graeme

Where do they live? Can I see them?

[Bob Torrance]

Combatentropy, I apologize in advance if the following message is terminology heavy but I wanted to provide a clearer picture why something more important is at play here and is not mentioned frequently, which Wavelet compression has taken advantage of very nicely.

Most Red users will be filming natural images. The power spectrum of natural images falls of more like 1/f^2, where f is the frequency, assuming that image statistics are at least wide sense stationary. For natural images, for which as mentioned, certain important image statistics fall off as square of image frequency, it will be helpful if we can capture this fall behavior using some ideally linear transformations. And, transforms such as Karhounen-Loeve / PCAs /etc., do capture this information very well, which is in accordance with certain Markov processes.

Wavelet processing is closely related with band-pass filtering, which electrical engineers have been doing for ages, and hence, such band-pass behavior ties in nicely with capturing the image power fall off, though, perhaps, not as ideally as KLT / PCA. However, it is reasonably good.

Therefore it is the properties/statistics of natural images that Wavelets is able to exploit more usefully than the often-used DCT/DFTs that are helping us here. On the other hand it is possible to provide example of images where Wavelets performs poorly because they could not exploit certain image statistics.

Oh, I see.........???

[Blair S. Paulsen]

I understand lp/mm measurements for lenses and the need for optical low pass filtering in front of imaging chips whether CCD or CMOS based. What I am curious about is the relationship between them. Is it as simple as using a lens that exceeds a particular lp/mm threshold in front of the OLPF to achieve the maximum resolution of the system?

Are there issues with the size of the photosites in relationship to the OLPF and lens combo? Are aliasing and moire issues linked to the OLPF and algorithm designs only or does the lens play into that equation as well?

Joofa raises some interesting points about the characteristics of reflected light in real world scenes and how the mathematical representations of that image information may play to the strengths of a particular codec. I have surmised, though cannot prove, that some of the RedOne firmware iterations addressed idiosyncratic "codec killer" waveforms.

[Blair S. Paulsen]

All of the serious D-cinema cameras can produce wonderful images with good lighting and quality lenses. I would pick the RedOne because it uses a large single plane sensor, is relatively small, lightweight and so flexible on set.

If you have seen the 4K reel that RED showed at NAB I think the image quality discussion stops with camera geeks like us. How about real world producer considerations. IMHO the RedOne is equal to or better than other options in these areas:

setups per day
handhold-ability with proper accessories
Steadicam friendliness
lens options
Depth of Field control
color information for grading
data rate to quality ratio
ready to edit

It is my contention that the remaining resistance to the RedOne in top end production environments is not about actual problems, its about perception and the security of sticking with what crews are used to.

[Graeme Nattress]

Where do they live? Can I see them?

Well, I found them when I journeyed to the planet Mysterium, where they were powerless because of the blue star that the planet orbited. Jim designed special sun glasses for the little Mysterium pixies, and hence they began to work for us. Jim also designed special glasses for me, so I could see the pixies and communicate with them, to get them to live inside our cameras.

Graeme

[Dj Joofa]

Well, I found them when I journeyed to the planet Mysterium, where they were powerless because of the blue star that the planet orbited. Jim designed special sun glasses for the little Mysterium pixies, and hence they began to work for us. Jim also designed special glasses for me, so I could see the pixies and communicate with them, to get them to live inside our cameras.

Graeme

Hahaha, that was really very funny indeed :) I really laughed reading it.

[Antoine Fabi]

...and the real reason for the fans inside the camera is to keep them comfortable? :)

...anyway they work pretty @#@# good.

Cheers.