Hello all,
Byron was shooting a project the night of the test results screening. He wasn’t able to attend and speak in person. However, Byron was able to type up the shoot notes into a brief synopsis which I found very helpful.
UPDATE – The test has been posted on Vimeo in web “HD” here, since the web version has compression artifacts, I’m going to see if I can get a quick 15-seconds or so clip of the test that I can make available for download so everyone can see the 3-cameras without web compression. This became Vimeo’s “Video of the Day” and got over 43K views in a day!
The complete and unedited shoot notes from Cinematographer Byron Shah are below this photo of Byron operating the Canon 5Dmk2 with a Hot Rod support system, and 1st AC Ethan McDonald operating a Bartech remote focus system.
———————————————————————-
Greetings Fellow Camera Peoples,
What a fantastic test this has been so far! Unfortunately I will not be able to be with you all for the premiere of GHOSTS as I’m shooting right now. GHOSTS is not your typical test. You’ll see no apples to apples comparison. And no color charts. What you will see is a short, impressionistic mood movie that intercuts the three formats in ways that highlight their strengths and weaknesses.
We shot each format with the desire to make the most it. The 3 minute film will play twice, once color corrected and once un-corrected with titles telling us what camera, pertinent settings, lens, filtration. After the movie plays twice I’ve included 2 shots from the Canon’s native format, H.264, 30fps, 1920×1080.
A little disclaimer: I’ve always loved little cameras. My first cameras were a Sony VX-1000 and a Bolex. I love hand held, quick camera placements, easy framing, and the pure (small) production style that small cameras usually mean. I’ve been hoping for a ‘digital Bolex’ for some time, but have found that small cameras usually mostly mean limited picture quality. I own a Canon 5D2, I use it every day, and simply love it as a still camera. This test was an attempt to see what it would be like to shoot a feature on these cameras.
A word on the post flow: In order to put the cameras on a level playing field, the footage from each camera was transcoded to ProRes422 at 23.98, 1920×1080, with the generous help and expertise of Tom Vice at FotoKem, a very good dude to know. The Lumix shoots 1920×1080 AVCHD natively at 23.98, but records in a 60 frame format, not unlike how a Varicam does it but also not exactly the same. Like the 5D2 this is an 8 bit format. This format does not playback, except via the camera, without being transcoded. Tom used Cineform to transcode the Lumix footage to ProResHQ, this he managed at 5 times realtime.
The Canon was transcoded from its native 30P 1920×1080 H.264 Codec to 23.98 ProRes (not HQ, an unfortunate oversight). This frame rate conversion Tom did twice, as the first attempt had stuttering and strobing issues. This conversion between 30 and 23.98 isn’t all that easy to do well. (Yes it can be done real time with a Terranex; or even on your laptop with Compressor, but not without causing ugly motion artifiacts and the clip length being altered and thus causing potential sync issues). Transcoding this footage took 20 times real time. Not really a feasible workflow.
RED we shot at 4K 16:9 on build 17 at 23.98 except where noted for high speed shots. Tom Vice transcoded to ProResHQ 1920×1080.
The RED and Ultraprimes and AKS were provided by Birns & Sawyer; the 5D2 and Canon lenses by Canon; the Lumix and PL Adapter and Rods by Hot Rod Cameras, the Zeiss ZF (Nikon F mount) and ZE (EOS mount) primes were provided by Zeiss.
And now on to the production of GHOSTS. We wanted this to be as real to true production needs as possible.
I cajoled my good friend Jason Ensler to direct this test. The actors and cameras were rarely not in motion. Because of our ambitious schedule and utterly stolen locations, most scenes were shot with all 3 cameras getting different coverage. The one aspect that was atypical of most productions was the absence of monitors. The Lumix has no video output. The Canon is HDMI out. Plugging in a monitor kills the LCD on the camera, and the VF is dead as well as the mirror is flipped up in LiveView mode, so to provide a client monitor from the Canon requires at least 2 monitors, one for the director and one for the operator. We could get our hands only on one HDMI monitor and it didn’t look very nice, so we lived without any monitors other than those on the cameras. This was fantastic on one level, as it meant no video village. On the other hand not all producers would go for this. And certainly the focus pullers would appreciate a good reference.
While shooting a few key differences were clear. The EOS and F mount lenses for the 5D2 look nice, however they are tricky to make functional in the field. The focus marks are pretty much useless on all these still lenses, the Canon primes and zooms, the Zeiss ZF and ZE primes, as well as my old Nikkors. The Canon primes are autofocus, so the barrel is actually a servo, so will spin around forever. We mounted a Bartec remote follow focus on the 5D2 in hopes of expanding the focus scale, but the RedRockMicro plastic clamp-on gears slipped a lot and rendered the system pretty flaky. Of course with permanently mounted focus rings on the stills lenses (see Hollywood Studio Rentals for this), and some time to prep marking discs for each lens, this would be mitigated. Judging focus on the LCD VF on the back of the Canon is tricky, especially when you can’t get your head right in there. And the optical VF is closed in live view mode as the mirror is flipped up. We didn’t have a Z-Finder (Zacuto’s Loupe finder) for this test, but of course this would help. W/O a remote follow focus you’d be forced to get eye-focus marks for each and every set up, which is tedious to say the least. We had some vexing donut issues with the 5D2’s setup (ie no donut small enough to black out the back of the mattebox) because the Canon primes especially have a very small front but a thicker barrel. Not insurmountable but annoying just the same. All this made switching lenses on the Canon very slow.
On the other hand the PL mount cameras, the Lumix GH1 and RED, had none of these issues. We used Zeiss Ultra Primes on the GH1 and the RED and Arri studio follow focus units (in addition to 300mm 2.8 Nikkor telephoto and a Angenieux HR 25-250). The marks lined up, the barrels were big, the UltraPrimes didn’t breathe, the matteboxes fit perfectly. The Lumix had Hot Rod Camera’s PL adapter, which is a surprisingly well built adapter for a camera that costs just $1500.
DOF on the 5D2 is very thin, which can be extraordinarily beautiful, but also makes pulling focus incredibly critical, even at deep stops (like T4 and T5.6). Again, very hard to judge critical focus on the camera.
The Lumix VF flips out as in a handycam, which made low angles really easy. This is a delightful feature. However, as noted, the Lumix has no video out.
Both the Lumix and Canon 5D2 were outfitted with Hot Rod Cameras Rod system, which feels nice and solid, but not designed to transfer weight to your shoulder. Good for speedy placements, quick moves between low and high angles. However bearing all that weight on your arms means it’s very difficult to hold long lenses steady. Here a shoulder mount system would be nice. None were available for us to test.
The small cameras are also WAY more sensitive than the RED, and are naturally fantastic at shooting available-light night exteriors. The Lumix was parked at 400 for most of the day (although the camera goes to 2500 i believe); at 400 is was easily as sensitive as the Canon at 2000 or so. The Canon we looked at from 50 to 2500. The RED of course is 160 (or call it 320 if you like).
FINAL COLOR
I was able to get one the best colorists in town to color this test, Arnold Ramm, who with Rob Walz, worked their magic for the first time on an Apple Color system provided by George Riskallah’s The Product Factory.
I’ll leave analysis of picture quality to you. Have fun.
Yours
Byron Shah
Another good GH1 review…We re on hte right track…
I went through these notes. No mention about the Lumix camera settings relating to film modes. Was this shot in smooth, standard or? How did they tweak those film modes? I notice you mention that it’s better to shoot with a lower kelvin, Why? Does this apply to the 3200K as well, like setting it for 2700k or so?
Thanks, The site is very informative and the gear your making is awesome!
Hey Illya, putting great information like this in such a frustrating web page design is like cruelty to animals! This little scrolly window thing is pointless and crazy-making. You can’t even see half the photo at one time!
We love what you’re doing, please make it easier to appreciate.
-Stu
My web guy has fixed the bug now. It just on me to go in and fix all my posts. My apologies. I usually get complements on layout/format of the site. I’m going to have to solicit some more feedback now to see if I need to re-design
Thanks for the kind words H. brusco. I have to look at what I wrote, but I’m pretty sure I said something around daylight Kelvin, not lower/warmer. Around 5000 kelvin seems the cleanest to me.
As for technical details, camera was in smooth, all zeros, no modifications to smooth.
wonderful product. I fear that this is just the beginning of a whole new realm of cheap digital filmmaking. Unfortunately for those of us who have been in this business for quite some time I suspect that products like yours sound our death knell. But they are really cool, really affordable and I really want one. About your website however, like Stu I believe you possibly could make it easier to read and easier to view the pictures. Just a suggestion.
Please keep up the good work and as new 4/3 and other compatible cameras come onto the market please provide the necessary hardware connections to make them work as well. I think it is absolutely tremendous that we have an inexpensive alternative to red and the D21, coupled with lenses like the new Zeiss set and the inexpensive (relatively at least) Cooke red set I think you are really on to a winner. Well done.
Thanks for the compliments Scotty. As someone who got started in this business in the ealy 90s, I’m going to disagree 100% about a death knell. I’ve heard that so many times before, most recently it was CGI, DV, and then HD.
Really, my product and others that have come before it are just tools to help people realize their cinematic vision. Amazingly, Super-8 is still kicking around, and I would have thought it would have been discontinued 10+ years ago. Instead new technology has emerged to make Super-8 better than it has ever been in the past (2K scans, anyone?)
I.
I guess we joined the business at about the same time then, perhaps I joined it slightly earlier. And having worked all over the world and still having only done a hand full of digital jobs in the past 10 years – I do stand by my comment about the death knell. Television advertising which has been the mainstay of the motion picture industry in many markets has changed. Products such as yours are tremendous equalisers in what has been up until now a very expensive industry to enter. The rule of thumb was too give television commercial work to people with experience, because the cost of producing television commercials was so high you didn’t want young creative people screwing up. Now of course, the cost of producing television commercials particularly in terms of cinematography and in postproduction has plummeted. The savvy producer could make five commercials for the price of the one they made 10 years ago and some now are. They can capitalise on the youthful intellectual optimism of the newly franchised and if truth be known their clients do benefit.
keep up the good work – for as far as I am concerned maintaining a visual reference like familiar 35mm (equivalent) depth of field characteristics and using good quality glass and manual focusing have really been the missing link between a filmic style and digital film making.
Cheers
I agree. The barrier to entry has never been lower. However, there is no substitute for talent (digital or analog). Your insightful comments remind me of some of the changes that occurred in publishing when “desktop publishing” exploded a couple decades ago now. There was a lot of concern that the industry (as the world knew it), would go away completely- that failed to happen. The recording studio industry too got scared when a top quality room could be build for only $30K, years before it was $100K, and before that is was $1M. Granted there was turmoil in both these industries. However, they were not “over” as prophesied. Talented people continued to thrive, no matter the technology.
On many big commercials the least expensive portion of the budget is already the 35mm film and camera rental. While some producers will never understand this and look for anyway possible to make something for less, the ones looking for quality and working at the highest levels already do.
I’m looking for a link to the test described above and am not seeing it. Maybe I’m squinting, but is the test posted online?
Link is added and Photo bug is fixed. Hopefully this won’t be a problem again.
I will say this is very exciting and thank you for doing these tests…..
I really would have liked to have seen the exact same shots with all three cameras trying to match each other…..that too me would have been a really good test to see how the three cameras stack up side by side
but that said this is a great test and I am shocked how good they look together….that said I am watching this on a little monitor on my computer…I think the real test is to watch it on the big screen or even on a big tv…..Have you done that and what do you feel is the final results. I don’t think we really got your final feelings words on the three different cameras.
but none the less good stuff!
Ron
This is such a great test. I have the GH1 and this along with your other info has helped me greatly. Are there any tips on getting the best image transcoded? I have been using log and tansfer>compressor pulldown removal>proresHQ – however it seems a little weak at times. I’ll see compression artifacts at time. Is cineform going to help on that you think?
Hi Justin,
I know that Liam and Frank (Editor of REJOUER) have been very happy with the CineForm workflow. I catch a compression artifact now and again, but I am extremely critical. My recommendation is to not preview the clips in the finder at any resolution other than full. That’s when I’ve noticed seeing artifacts the most, the good news is that these artifacts you see in preview, are existing only on this scaled playback and not actually in the images themselves. Check the images on a big screen a full resolution for confidence in the transcoding method. If you see artifacts there, they are in the captured image.
Red One, Canon 5D MkII, and Panasonic Lumix GH1 Side by Side
http://www.youtube.com/watch?v=-lqUxnI47tE
I didn’t know it hit YouTube also, thanks for the link. YouTube’s “1080p” for this is surprisingly good.
I.
Illya Friedman
President
Hot Rod Cameras