The Gentleman's Underwater Bubble Blower Thread

[Jeff Whitehurst]

Much to say on this subject I have, for now however, there's a great interview with Mr. Johnny Friday on UWC, called "Adrift in a Sea of Pixels"

www.UnderwaterCinema.com

Nice Rudi and Johnny thatʻs an excellent resource for any shooter! I would add an easy but very helpful step I have adopted, which is to create a PDF and HTML reference of the footage using the Foolcat R3D cataloging software. It makes finding and retrieving shots much easier.
www.foolcolor.net
Cheers, Jeff

 

[Greg F]

Personally I experience a lot of issues with the red and green channels.
It is very difficult to shoot ambient with the Epic Mysterium-X whereas a simple D7000 shoots perfect images in the same conditions.
I talked a lot with Rudi about this and he says there are quite some Epic owners whom experience the same issue.

Where I spend the bulk of my time there is some pretty substantial color issues. I am still unsure how to deal with it

 

[Rudi Herbert]

Epic presents a conundrum of sorts when compared to the R1 underwater, most owners of both seem to agree the R1 yielded better, more predictable results, if with less resolution. In terms of color accuracy and precision, Epic has really been a step backwards, which keeping in mind the R1 was a step backwards when compared to 3 chip designs, well, has left us with images with lots of resolution but nowhere near the color we need. Being that the sensor is the same on both R1 and Epic, then yes, the OLPF could be the culprit, but I'd venture a guess that Redcode processes Epic footage differently than it did R1's, with the sensor now being used to its maximum resolution, maybe the extrapolation algorithms are different? There is always inordinate amounts of crosstalk between blue and red channels in Epic underwater footage, and under certain conditions it becomes immense, impossible to fix really. Certain things help bad looking footage quite a bit, a big one being reverting back to Redcolor/Redgamma instead of RC3/RG3, which restores more natural looking images, but still there are cases where the footage is, truly, unusable. I have never had to spend so much color correcting any other footage as I do with R3Ds, and never have the results been so unpredictable, non repeatable and unsatisfying.Case in point. After carefully studying some really bad footage shot on Epic under normal conditions in very clean waters by one of our members here, I even advised him to return his camera to RED, from where it eventually came bacck without any flaws having been found, but I can assure you all that had I been on assignment for a client and that's the footage I would have delivered, you can be certain I would have never been hired again.

So I skipped the Epic altogether and stayed with the R1, but after having seen footage shot first by the F65 and just a few days ago by the F55, I am waiting for Dragon to make a decision about future image systems purchases. Both F cameras exhibit way better colorimetry underwater than Epic and R1, so unless Dragon brings something substantially different to the table, then decision time for me will be, shall we say, interesting. I am invested in RED both as a technology and a company, but ultimately, I don't want, and can't afford to, do disservice to myself by acquiring images with a system that is not the best underwater. And for me, color fidelity will trump all else underwater. In fact, even if Dragon delivers as we all hope that it will, I think I'm going to wait until F65s are selling in the $20Ks on Ebay and snap one off, those things make wickedly beautiful colors.

 

[Greg F]

Hi Rudi, I do agree that using Redcolor/Redgamma does fix immense amounts of problems... This lends credit that its possibly the color science? I love the images it outputs and when under artificial light its MAGIC.

How many people here are shooting ambient blue water at 25+m for extended periods of time? This is where I seem to find my color issues the most. Ambient light is my main thing that I am using underwater unless sharks get inside what my buddy and I call the "Green Zone" where they turn green from our lights before they hit full color. Its not uncommon for them to get inside that green zone however when they are not I am left with ambient. Thanks for bringing up these points Rudi I really appreciate it. My dive buddy who shoots with a HR520 Sony Handycam that is somewhat older now behind the fathom port with gates housing seems to have way better color out of the box. Infact his color is so nice by comparison to mine that he has complained about my raw footage that I have not corrected or not been able to correct and asked whats wrong with it. Its pretty dissatisfying but I am willing to tolerate it as I am slowly figuring out how to pull blues back and nuke the Magenta but its always guesswork it seems? Is this what you refer to as non predictable results Rudi?

Its very complicated depending what I do with the images and all of this I am doing is trial and error as I really have no clue what the hell I am doing to try to fix this LOL...

Edit:

And to Johnny, Thanks for ruining my world in that interview.. You made me need to sort my 15tb of video/stills

 

[Rudi Herbert]

Greg,

Yes, by unpredictable I mean that even in somewhat similar situations, sometimes even in clips shot one after the other, you may encounter noticeably different color renditions out of the Epic. To the point where batch processing clips is a risky endeavor, what may one looks good makes another not so good and yet another one terrible. I am, of course, talking about clips shot on the same dive, same conditions, basically same light.

I think the "magic" that comes with RC3/RG3 or even RedLogFilm is one of complex exchange between the red and blue channels, where the low end or shadows of the blue channel is reconstructed (for lack of a better term) from the red channel, which is very clean on the RED cameras, and then in exchange, the mid to high midotnes on the red channel take some info from the blue channel, which is where the blue is at its strongest (underwater). I've come across this conclusion by isolating noise and degradation patterns on both blue and red channels and finding lots of similarities, which are absent in the green channel. I could of course be wrong, of course, but RED have always been very mum about this and understandably so. I believe the R1 with RC/RG had a much more primitive, i.e; natural, way of recording color information and thus its colors were much more solid. Consequently, if you process R1 footage through RC3/RG3 or RLF, you will get the same anomalies the Epic exhibits, and likewise, Epic footage through RC/RG behaves much better albeit at lower DR. Another way I've found to clean up some of the chroma noise the Epic generates, specially when shooting ambient in deep water is to use negative values in Alchemy in RCX, to the tune of -300 to -700, which reduces localized contrast an eliminates a lot of the nasty artifacts though not all. You will then have to increase local contrast in a different way, which you can do in a lot of applications, my favorite being Color Finesse.

Right now, the Epic/Scarlet is a winning combination for us underwater shooters in terms of size, RAW and price, but its colors lack severely when compared to other cameras. This is not a dig, is the truth and something I hope gets solved with the new sensor and new color science. Respectfully, while Jim and Co. may believe their most valuable asset is Redcode, and I agree that a camera like the R1 could never have happened without a way to record compressed 4K RAW, I feel in terms of color quality it is the biggest liability of the RED. None of those problems are a consideration when shooting on land, but in the color/tint skewed underwater world, either Redcode is doing us a disservice, or the MX sensor is at fault or a combination of both. I love the RED and appreciate every other feature it brings to the table, but one more time, as far as I'm concerned color rendition trumps everything else underwater.

 

[Greg F]

Do not confuse my complaints with not liking the camera, I absolutely love what it does. But yes I notice that there is very odd variance.. I never batch process I have to do it all by hand otherwise it gets very odd.. I can batch process stuff when I use artificial light.

Its just some oddities I have to work with and learn how to deal with that are sometimes irritating. I have yet to do any deeper stuff I am hoping it does not get any worse if I go deeper at Molokini for instance and drop down there.

 

[Pawel Achtel]

I have not experienced any problems with colours in Epic footage. I'm not saying there is no problem, but can you guys demonstrate what you think the problem is?
Maybe post sample r3d frames?

We have done a lot of R&D in underwater colour science and I'm currently shooting with highly filtered and modified optical path with custom colour grading to match, so I may not be experiencing what others are experiencing.

 

[Frazier Nivens]

Oceanic White Tip Shark, shot today

Oceanic White Tip Shark, shot today

Fun time today filming Oceanic White Tips. Had 6 beautiful creatures up close all day today. Here for a bit still, quick frame grab from EPIC. Crazy good day, beautiful blue water.

 

[Rudi Herbert]

Pawel,

Retrace the thread and you'll find a few samples posted by people who got odd looking footage they could not fix. John Shaw was one of them with some whale footage I think, and there was also a similar thread about whales under the Epic sub-forum, some Nat Geo job, which you commented on if I'm not mistaken. I've also been sent privately similar footage from four other posters on this thread, including someone whose footage was so bad all the time he had the camera sent back to RED and they decided there was nothing wrong with it. Not all cameras exhibit this problem, and some that do don't do it all the time, while others do, but in general, I have ascertained some patterns that seem to repeat:

- R1 footage did not exhibit any of these problems ever, both M and MX.
- The majority of the problems exhibited by the Epic have to do with magenta tinted midtones and green tinted shadows. Interestingly, while too much or too little green (magenta) might seem a problem with the green channel, in fact, it is crosstalk between red and blue channels that creates the problem. I am almost certain of this.
- The problems usually happen in deep water with ambient light, and even minor variations in the quality of the light, i.e. sunny or overcast skies, make a difference in the severity of the problem.
- However, the problems also have happened in very shallow water with plenty of light, and in very clear water locations, which is baffling.
- When trying to fix the problem, if you mkanage to restore tints and tones where they belong, one constant byproduct is terrible red chroma noise all over the middle range, with actual red speckles all over the place, most of all in the blue water. This then needs several denoising passes which usually degrade the footage badly, many a time unnaceptably.
- Throwing a little of artificial day balanced light into your frame, anywhere where the sensor can actually perceive it seems to go a long way towards fixing the problem...
- Debayering with Redcolor and Redgamma fixes many of the unnatural tints considerably, water becomes blue again, but there is very little information on the warm areas after that, so you gain a more normal looking image but are still left with very little color to work with.
- I've attempted to fix the bad footage not only in the regular suspects AE, Da Vinci, etc but also on a no-expenses-spared Pablo 4K suite with a very well known Hollywood colorist. Results are the same no matter.

This is NOT a result of improper shooting, many people seem to doubt themselves and blame their lack of experience with a professional system like a RED for these subpar images. Not so. Most of the images I've received for analysis were properly exposed and shot, with really good histograms and no lenses pointed into the sun or whatnot. RED is truly like film, proper exposure and focus are all you need to worry about, exposure depending on your aperture since your shutter is set, and focus is not even a factor in the quality of the images, only on whether they are sharp or not. So if any of us is out there shooting together with, say, Howard Hall, and both have the same system, the same exposure and are shooting from the same place with the same framing, then the images would look identical. Magenta looking water has nothing to do with whether we are casual shooters or a renowned professional.

 

[Michael Maes]

Howard's Epic is magenta-itchy too. (I hope he doesn't mind me telling this here :undecided
Mine is half a disaster: exposure needs to be spot on and, if deeper than 10 feet in tropical water, it is almost impossible to shoot ambient.
I love the Epic but, frankly, in this regard, my 'silly' D7000 shoots better.
I truly hope Dragon will fix this issue because else I will get over with it and migrate to Sony (I think)....

 

[Michael Maes]

Much to say on this subject I have, for now however, there's a great interview with Mr. Johnny Friday on UWC, called "Adrift in a Sea of Pixels"

www.UnderwaterCinema.com

Great work Johnny and Rudi. Nice to see this subject so elaborated

 

[Pawel Achtel]

Rudi,

I have R3Ds from Jon and had an opportunity toplay with them.

While I was able to get decent results, what caught my attention was rather low span of contrast range in all R, G and B channels. Essentially lack of blacks and lack of highlights. Kind of milky looking footage and when you want to recover your reds, it ends up magenta poo, for lack of better description

This makes it hard to colour correct using traditional approaches.

However, I'm not so sure if this is caused by the camera, the actual lighting or the optics. The loss of contrast can be caused by many factors, most likely associated with water clarity (do not confuse this with visibility), ambient light scattering from particles, or optics in general from dome, lens, right to the OLPF and sensor well area.

Ambient properties can cause some dramatic contrast changes depending on light direction, angle, intensity, sediment, etc...

It would be interesting to figure out what is causing it.

As a scientist, I'm intrigued

 

[Michael Maes]

Well again, i think this is a decision to be made based on actual working results. I talked to Howard about this and he really likes the PC with speed and upgradability and i can't agree more with him. Howard is one of the smart ones. That said, i think waiting for what apple comes out with is a wise decision. There's a lot more to this decision than just the speed of which computer-PC or Mac. As mentioned, what sort of investment does one have in software and hardware for either PC or Mac. That's currently where i stand. Also, i work in both Premier, After Effects as well as FCPx and find myself really liking FCPx for basic stock work. I think until one really sits down for a few weeks, you cannot make any determination based on these threads on how good one or the other is Pr or FCPx....Hell, i love After Effects and Premier...but use those when i need them. I think when apple upgrades, we'll see a new chipset that will fit our needs. But doubt we'll ever have the ease of upgrading like PC's....But i''ve now become a fan of FCPx and only adopted it about a month or two ago...otherwise i was full time in Pr/Ae. But having ability to use both sure is nice, so we can only do that on a mac.
I guess having both would be a luxury, but for most of us one man bands, not really practical. Oh...now let me mention using my Retina display with Magma chasis and Red Rocket card...that thing files better than my 2009 macpro loaded to the hilt with RAM, Quadro 4000 card etc.....

I would be careful of throwing Apple under the bus at this point. What's missing with both of these Programs: FCPx and Adobe that is really critical is Batch Processing that FCP7 has....that's a killer NOT having it. Also, a way of using the smart folders and keywords in FCPx to actually make those folders or keywords into working Metadata nailed to the actual asset and not just within the program-- that would sure be a winner.

Howard i think is committed to going PC after talking with him, but we plan to do some side by side tests to see how each pans out. With luck he'll hold off till Apple releases their new MacPro...but i tend to feel him leaning towards a new solution NOW...i think he's had his machine for over 6 or 7 years. So he's just dying for more speed.

...As for the epic underwater...well, i think all of us see a bid difference. I was happier with my red1 images underwater....i'm hoping for a better color science. I wonder if the problems are due to the OLPF of the Epic.

I just heard from "a guy who knows a guy" who works in the R&D and the new MP will be (if it's true of course) a small form factor with no PCIe internally but with 6 thunderbolts which you then connect to external Apple chassis. Kind of like a Lego-kit ;-)
Well, even if only 1% of it is true, it sure reassures me that Apple is überhaupt working on a new Mac Pro and it will be a totally rethought system!

I think Howard wants to switch to PC because he doesn't trust Apple anymore. And truth told: what reassurance has Apple given its professional clientele?
What transparency?
Don't get me wrong: I love Apple

I never shot the R1 but what Rudi showed me of an R1 compared to my Epic.... it's some serious steps back (underwater).
Once again: I'm hoping Dragon will deal with this!

Now I'm going to take your input and see how I can improve my storage and labeling ))
Thanks Johnny!

 

[Johnny Friday]

To be fair, has apple ever been forthcoming about their future plans? I think that's just how they are and in the end i think they still surprise us. I dropped FCP i think about 3 years ago and was Premier all the way and as my mileage got worse on my old 2009 macpro i also am considering a pc. BUT, I've also heard of the all in one box....and expect apple will be to market late as usual. But as soon as i learned premier and felt comfortable...along came FCPx and have to say, i'm impressed. We'll soon see though....I'm holding off on making any big purchases until Apple speaks....all i have to do is remember my usual windows experiences with wiping drives and reloading software on a regular basis due to either virus or just weird windows OS issues. Also, i like FCPx and can't play with it on a pc. If it comes to it and apple comes up a dud, then I'll be right there in PC land with Rudi and countless others. That said, my personal computer will always remain a mac though.

....as to R1 and Epic and color science etc....Well, i'm processing a bunch of clips for a client and all my R1 footage is far easier on the eyes and takes far less post work to push out good looking images. As for epic and the big magenta hits, there are definitely ways to deal with them in the individual color channels, but it is a let down how footage in epic vs. R1 resolves. And if pushing hard in color work, it really helps to have a proper I/O device to view your images in the color space you intend to deliver them to. And neither of these NLE's---FCPx or Premier can hold a candle to working in Resolve for getting into those problem images. But if the images has good DR and exposed well, the tools in Premier and FCP and/or plugins do a great job. As Rudi mentioned, Color Finesse does great job...i find it and almost every other plugin on my Mac an anchor slowing me down.

 

[Simon Christidis]

Nice work Rudi

 

[Greg F]

Ambient properties can cause some dramatic contrast changes depending on light direction, angle, intensity, sediment, etc...

It would be interesting to figure out what is causing it.

As a scientist, I'm intrigued

Well I have footage of me spinning around all directions, and by spinning I mean recording while I am doing a twirl underwater quite extensively. I recorded that over and over as whales were making noise and no sharks. I have rolled over shot into the sun with sharks above me obscuring sun and not obscuring sun it is very very problematic. I am unsure how I could even get the footage to you unless I mailed it on something? But the reality is this is a very real problem.

And yes I shoot in visiblity that is very good. I shoot in water when I set at 30m I can see the surface clear as day, I should take a trip out to Molokini where the clearest water I have ever seen, I have yet to go out there with my Epic.

 

[Pawel Achtel]

Greg,

In your example, did you find the problem when shooting in all directions or just in some directions?

Was any of the footage good on that dive?

 

[Rudi Herbert]

All,

Further to my post, a few more observations regarding this issue:

- it is not a matter of compression, Epic foootage at 3:1 looks better and it is easier to tweak but the abnormal tints are still there.
- water clarity is not a determining factor, I've shot proper looking images in dirty water with lots of particles and seen bad images from some of the cleanest, clearest locations on earth.
- optics are not determining, color casts of that nature can only happen at sensor/software conversion level.
- contrary to logic, the blue channel is the one with the biggest problems, even if there is nothing but blue light underwater. So the more "perfectly" blue the surrounding water is the lesser the problems. Cyan and green water exacerbate the issue, deep blue water from bottomless places results in lots of red chroma noise, whereas typical blue water with less amounts of green and red yields better results. the best looking Epic water I've seen are from the Bahamas or the American Pacific, both South and North, with stuff in Red Sea looking terrible, followed by the Austalasian Pacific. Caribbean stuff is a toss, some may look great and some may not. This is all VERY subjective and I could be smearing myself in manure by saying this but the precise wavelenght of the water does make a difference in my opinion.
- Shooting ambient, camera orientation relative to the sun is important. Sun behind the shoulder: bluer water, sun in front of the lens: (even at proper angles with no light entering the lens directly) magenta water. Shooting downwards: bluer water but lots of noise, shooting horizontal: less noise but water rendition highly irregular, with shafts or columns of water of markedly different colors in the same frame AND lots of macro blocking and nasty artifacting in the areas of confluence between high light and shadows.
- Shooting at anything over F8, regardless of the lens, results in diffraction limited images that have very low contrast and tons of artifacts. In those cases, the images may or may not exhibit the abnormal tints, but the limiting factor will be the lack of contrast, which is IMPOSSIBLE to fix in post. I strongly recommend using screw on ND filters or even gel taped over the sensor for those shooting in high light, shallow water, like those shooting whales and the like in the Pacific or places like that. (of course this does not apply to macro shooters with lights) The sweet spot for the MX pixel characteristics and light wavelength seems to be at F4-F5.6, I consistently get the better color rendition at those settings.
- Certain lenses exhibit very strong color fringing on the MX sensor, which they do not exhibit on other cameras using the same dome ports and port-lens alignment. Those other cameras are also S35 sensor sized. I have observed this phenomena on the Canon 8-16 fisheye the most, sometimes on the Tokina 11-16, plenty on the Sigma 8-16 rectilinear and very little on the nikon wide angle primes and zooms. The Tokina 10-17 fisheye has little color fringing but its contrast is the worst of the bunch, though this I've seen on the other cameras as well. The best images I've seen on the Epic and the 8"dome combo are usually from 14mm rectilinear and 15mm fisheye primes. The Nikon 14mm F2.8 prime looks stunning, then so it does on every other camera. The lowly Sigma 15mm fisheye is the best looking fisheye underwater for my money.

I think I should stop now. I am trapped in a difficult situation here. I respect what RED has done and the doors that undeniably it has opened for all of us. I, for one, had gotten so fed up with the ownership experience after my two F-900s that I vowed never to get back into the big sensor, big camera game ever again. the R1 helped me back in this world and I am eternally thankful for that. But to be honest, as we stand right now, I DO NOT recommend the Epic for underwater work. The R1? Sure, better option than BMCC and all other similar 4K stuff out there, except the Sony F line, down to the F3, which look way better underwater even with their low bandwidth codecs. I am praying for a miracle with the Dragon, I really am.

 

[Tom Hamilton]

Very much appreciate you sharing your experience Rudi, lots to think about there, including the expensive Dragon upgrade, hopefully you can get your hands on some underwater Dragon footy and let us know.

Also wanted to now address an issue I promised to a few weeks ago regards the Amphibico pistol grips, here is what I received from the factory engineers today;
-----------------------------------------------
COMPLETE REDESIGN OF AMPHIBI-GRIPS


Since the Aquatica takeover of the Amphibico brand, a major redesign has been implemented to the Amphibi-grip system, now in use on all our housings, including the New ROUGE for the Epic & Scarlet digital cinema cameras.


Prior grips were made of plastic, temperature changes caused the material to expand and contract, potentially compromising the integrity of the grip. Here are the key pistol grip redesign features that now make the Amphibi-grip a solid, trouble-free component of the Amphibico line of video housings.


 Aluminum construction, anodized and powder-coated


 Grip insert is fully potted, for a redundant, water-tight seal


 Grip electronics are fully conformal-coated to protect against moisture, dust, chemicals, and temperature extremes.


 The grip trigger has been redesigned and fitted separately into the grip to insure fail-proof performance.


 The grip is rigorously pressure tested before and after installation into the housing, to a depth of 330ft (100 Meters)
-----------------------------------------------

So, hopefully that should lay to rest any further concerns out there, please pm me if anyone would like more information. The Rouge is selling briskly, orders are keeping the line humming, dozens of units now out there in the wild, some of them with very well-known shooters. No adverse reports as of yet, none expected by the very talented engineering team that designed the Rouge.

Warm regards to all,

Tom

 

[Michael Maes]

Very much appreciate you sharing your experience Rudi, lots to think about there, including the expensive Dragon upgrade, hopefully you can get your hands on some underwater Dragon footy and let us know.

I'll get you 6K footage Rudi as soon as my Epic is upgraded.
Cheers!