PDA

View Full Version : Scarlet Dragon 150fps @2k WS... is it even worth it? You decide.



Christopher-lee dos Santo
11-13-2014, 09:40 AM
So... as the title suggests. Your thoughts gentlemen? Is the option to film 2k WS @ 150fps worth it? I say this because I've heard that the 2k image from the 6k sensor is grainy/noisy as the sensor crops in. (Please don't quote me on this, just posting what I have heard.) The reason I started this thread was to open a discussion based on Scarlet Dragon being able to shoot 100fps and up at higher resolutions compared to the old Scarlet MX. With Dragon and it's greater redcode compression, 150fps seems like a damn good deal to some people, even at 2k WS at redcode 7:1. So, is shooting at 100fps at 3k and 150fps at 2k any good for lets say a music video that will finish up in 1080p? Or a feature that will be delivered at 2k? I believe most features still get released at 2k on Cinema with DCP. So I guess another good question is whether shooting dragon at 3k/2k will deliver a good enough image for a feature being delivered at 2k? This is not a Scarlet vs Epic debate, but rather, an informative thread on the real uses for Scarlet Dragon 100-150fps. That being said, I have shot some tests here at my home. The tests were done at different frame rates and speeds on the Scarlet Dragon with a crappy 18-135mm stills lens. I didn't have time to pull out the Zeiss primes and test thoroughly, but I will post it up here after I have cut it together. In the interim, feel free to post any findings or opinions you guys may have. I recently finished shooting a feature film on the Scarlet Dragon at 5k WS 7:1 23.98fps here in South Africa, and the images are breathtaking with the Zeiss lenses. So I know for a fact that the scarlet Dragon is a damn good camera for that application.

EDIT - TEST FOOTAGE ON PAGE 2 !!!!!!!!!!!!!!!!!!!!! AS WELL AS FINDINGS AND SCREEN GRABS FOR CROP SIZING ON PAGE 3!!!!

Alan Peterson
11-13-2014, 09:55 AM
I can't speak for Dragon, but 3K makes for a very nice HD finish on the MX sensor. That provides enough resolution to throw a bit away for stabilization/reframing if necessary (which should be much less of an issue for a music video than when trying to get extra reach out of a super telephoto lens) and still oversample your final output by 20% or more.

Youtube compression is killing some of the detail here, but this was shot at 3K on MX, cropped to 2.5K in post, then encoded to 1080p
https://www.youtube.com/watch?v=CDpSt09IOYs

Akin A
11-13-2014, 11:58 AM
You might find some helpful info in this thread:
http://www.reduser.net/forum/showthread.php?123009-Scarlet-Dragon-Resolution-vs-compression-decisions

Ryan Hamblin
11-13-2014, 12:11 PM
I think as long as you stay at 3k you won't have an issue delivering 4k. We have scaled up to 4k on many projects and with a bit of sharpening your hard pressed to notice a difference unless your really pixel peeping it

Nick Morrison
11-13-2014, 12:17 PM
I've shot 2K on MX and made it work. If you keep your compression as LOW as possible, 3K and 2K certainly become more usable. Everything has it's place. I'm pretty sure this was 2K on a Zeiss 10-100:


https://vimeo.com/81559139

Christopher-lee dos Santo
11-14-2014, 04:12 AM
After some tests. It seems at resolutions of 2k and 3k, lens choice is a very important thing to keep in mind. The footage is soft, and if iso is above 1600, expect to see noise. REALLY BIG NOISE.

I think I need to conduct some tests with the CP II Primes as opposed to this cheap stills lens. The video I will upload will have some 5k 7:1 footage at 23.98fps and 5k 16:1 at 60fps combined with 2k 150fps @7:1 and 3k 100fps at 10:1.

Currently, I am not impressed in the 2k and 3k samples thus far. Compared to the 5k stuff, it's almost like I am shooting with another camera. That being said, to be fair, I was using zeiss primes in a controlled set environment on the 5k 23.98 and 60fps footage. I will conduct more tests, but as a side note, 5k shot with the cheap still lens looks terrible. So I may chalk up the reason for 2k and 3k looking really bad due to a lens that will never ever touch my RED again.

I guess we can learn from this. Any indie guys who want to buy a Scarlet and use cheap stills glass, be warned, you get what you pay for.

More to come. I will post some stills in the interim.

sam karr
11-14-2014, 04:16 AM
After some tests. It seems at resolutions of 2k and 3k, lens choice is a very important thing to keep in mind. The footage is soft, and if iso is above 1600, expect to see noise. REALLY BIG NOISE.

I think I need to conduct some tests with the CP II Primes as opposed to this cheap stills lens. The video I will upload will have some 5k 7:1 footage at 23.98fps and 5k 16:1 at 60fps combined with 2k 150fps @7:1 and 3k 100fps at 10:1.

Currently, I am not impressed in the 2k and 3k samples thus far. Compared to the 5k stuff, it's almost like I am shooting with another camera. That being said, to be fair, I was using zeiss primes in a controlled set environment on the 5k 23.98 and 60fps footage. I will conduct more tests, but as a side note, 5k shot with the cheap still lens looks terrible. So I may chalk up the reason for 2k and 3k looking really bad due to a lens that will never ever touch my RED again.

I guess we can learn from this. Any indie guys who want to buy a Scarlet and use cheap stills glass, be warned, you get what you pay for.

More to come. I will post some stills in the interim.

What cheap still lenses were they?

Christopher-lee dos Santo
11-14-2014, 06:29 AM
I was lazy and just grabbed a Canon EFS 18-135 kit lens that was nearby. Colour rendition and sharpness are definitely compromised on said lens. Also, the Camera did not recognize the lens and wanted to calibrate/make a profile for it. Could also be the reason why the images just looked bad. I'm not exaggerating when I say my old 7D would shoot a cleaner 1080p image than what I was seeing. Or have I just been spoilt because of the Zeiss Primes?

I am doing some more tests today. This has got me interested. I'm going to try out a Canon EF 50mm F1.4 and the F1.2 L version. Then the 70-200mm L lens. Will cut the tests together with the Zeiss CP II's from a controlled shoot and compare 2k, 3k and 5k variable frame rates.

Nick Morrison
11-14-2014, 06:39 AM
I was lazy and just grabbed a Canon EFS 18-135 kit lens that was nearby. Colour rendition and sharpness are definitely compromised on said lens. Also, the Camera did not recognize the lens and wanted to calibrate/make a profile for it. Could also be the reason why the images just looked bad. I'm not exaggerating when I say my old 7D would shoot a cleaner 1080p image than what I was seeing. Or have I just been spoilt because of the Zeiss Primes?

I am doing some more tests today. This has got me interested. I'm going to try out a Canon EF 50mm F1.4 and the F1.2 L version. Then the 70-200mm L lens. Will cut the tests together with the Zeiss CP II's from a controlled shoot and compare 2k, 3k and 5k variable frame rates.

Couple of things about shooting 3K and 2K. This is how I think about it anyway. You HAVE to compensate for the lower resolution (3K, 2K) by OVERCOMPENSATING in ISO and COMPRESSION. Ie shoot 320 and as low compression as possible.

As I see it, it's basic math. You are starving the image of resolution by shooting 3K/2K, so you need to make sure WHAT you shoot is as CLEAN as possible. Hence feed it LOTS OF LIGHT, shoot LOW ISO, and as UNCOMPRESSED as possible. Any signal/noise issues are EXACERBATED at 3K/2K because you are punched in on the pixels.

Shooting 1600 ISO is very unlikely to deliver good results, probably even on Dragon.

And yes, LENS choice is important. Because you are taking a resolution hit, try and shoot a SHARP as possible...ie stopped down to 2.8/F4.

RED presumbaly makes 3K/2K available because it's a TOOL, like any other. But it's a tool that needs to be used properly for optimal results.

I've shot LOTS of 3K/2K MX footage that looked great, but always when I followed the above principles.

Akin A
11-14-2014, 09:22 AM
As I asked in the other thread, do you guys think an "unsoftened" OLPF option would improve 3K/2K HFR even though it would introduce moire?

Michael Tiemann
11-14-2014, 10:05 AM
As I asked in the other thread, do you guys think an "unsoftened" OLPF option would improve 3K/2K HFR even though it would introduce moire?

Moire is a freakin nightmare on the BMCC. I've basically abandoned that platform in favor of RED and GH4.

Christopher-lee dos Santo
11-14-2014, 02:58 PM
As I asked in the other thread, do you guys think an "unsoftened" OLPF option would improve 3K/2K HFR even though it would introduce moire?

I've done some more tests. It was definitely the lens that caused the soft look and artifacts. That 18-135mm is not a great lens to shoot RED with. I'd never use it professionally, that is what the Zeiss are for, but I was really surprised in the vast difference in quality. The colours are milky and it's almost as if I am shooting with a 4:2:0 camera. Also, soft as hell.

I tried the little 50mm F1.4 and man, what a difference. Such a little dynamite lens. 2k and 3k seem usable when exposed correctly and in focus. I will cut the stuff together and upload it for you guys to have a look at my comparison.

There will be various frame rates used with Zeiss CP 2 lenses, the 70-200mm F2.8 L glass from Canon, the 50mm F1.4 and finally the crappy 18-135mm. You can actually see what a difference lenses make at these resolutions.

Stay tuned. Video to come as soon as my slow internet uploads it.

Larry Cheuk
11-14-2014, 11:44 PM
Glad you posted this Christopher. I did a ton of tests with this this week after I got my upgrade back. Just like everyone mentioned, the lens and low compression ratio made a big difference. I was using my 24-70L stopped down to f4.5 and it was fairly sharp.

Also blackshading is totally critical for the higher frame rates. I'm assuming most people would use the 2K and 3K primarily for the higher frame rates. I forgot to blackshade during one set of tests and after I did, it was unbelievably better in terms of noise.

Anton Shavlik
11-15-2014, 12:55 AM
I've posted my own test before, and found basically the same thing. You need a sharp lens, lots of light, and low compression and it'll look acceptable. It's great that red allows us to crop to this format and run it at higher speeds / lower compression even though the image quality is not on par with what we expect to see from a red camera. I would say it still looks better than a barebones FS700 can capture, and it's in the r3d format which allows a lot of cleanup in post.

Craig Lees
11-15-2014, 04:31 AM
Shutter speed makes a difference in sharpness too. And yes, that canon 18-135 is uber soft, even on other cams, especially when shooting high speed when you can rarely shoot at f8.

Cheers,
Craig Lees
Epic X
UK

Christopher-lee dos Santo
11-15-2014, 05:34 AM
Here she is...


https://vimeo.com/111890797

Adam Montville
11-15-2014, 11:42 AM
Wow, those Zeiss lenses really are something!

Akin A
11-15-2014, 05:08 PM
Thanks for doing that test, Chris. You mixed the shots so well that the quality drop wasn't noticeable, which is probably how an actual 2K finished project would work.

I plan to do some tests soon on a close-up w/ skin tones to compare the same shot in a narrative framing, using Sigma Art lenses.

Christopher-lee dos Santo
11-16-2014, 03:47 AM
Thanks for doing that test, Chris. You mixed the shots so well that the quality drop wasn't noticeable, which is probably how an actual 2K finished project would work.

I plan to do some tests soon on a close-up w/ skin tones to compare the same shot in a narrative framing, using Sigma Art lenses.

My absolute pleasure. Thank you. I remember when I was doing research on the Scarlet Dragon, there were literally no clips or test footage from the camera on the net. So, I hope this can help a potential buyer in the future and bring them toward the RED family. Vimeo still adds some compression to the video, so maybe that hides the difference well. Also, I shot mostly long lens with shallow depth of field (ND 9 and 6 on the matte box) with the tests at home, I did some wide open and some around F4-F5. The Zeiss shots were also ND 9 with Grad, and around the same T stop. In the full 2k file, you can definitely notice a difference between the shots, probably because the Zeiss lenses being so sharp and their colour rendition being amazing. That 18-135mm lens is awful in comparison. The colours are milky and its alround soft. I would imaging one would need to use A.D.D or unsharp mask on the shots.

I look forward to seeing your tests.

Larry Cheuk
11-16-2014, 01:11 PM
Really excellent work Christopher! With the Zeiss 2K shots what were the T stops you were using? You mentioned they are around the same T stop.

Christopher-lee dos Santo
11-16-2014, 04:16 PM
Hi Larry, if I recall they were between T4-T5. Unfortunately the Zeiss lenses do not speak to the camera, so it is not saved in the MetaData. I'm uploading some stills and you can see the difference between 5k, 3k and 2k on the same shot of the bicycle wheel. These are all shot with the Canon 50mm F1.4. This will give you an idea of sharpness and the size of cropping when changing resolutions.

For the full resolution -- Go here http://s853.photobucket.com/user/Crash_Nemesis/media/RED%20Dragon%20Tests/A023_C023_1114JG0000428_zps2c2a47fe.jpg.html and then select the magnifying glass to see the proper full 5k image.

5k
http://i853.photobucket.com/albums/ab97/Crash_Nemesis/RED%20Dragon%20Tests/A023_C023_1114JG0000504_zps222c1ecd.jpg (http://s853.photobucket.com/user/Crash_Nemesis/media/RED%20Dragon%20Tests/A023_C023_1114JG0000504_zps222c1ecd.jpg.html)

http://i853.photobucket.com/albums/ab97/Crash_Nemesis/RED%20Dragon%20Tests/A023_C023_1114JG0000428_zps2c2a47fe.jpg (http://s853.photobucket.com/user/Crash_Nemesis/media/RED%20Dragon%20Tests/A023_C023_1114JG0000428_zps2c2a47fe.jpg.html)

3K
http://i853.photobucket.com/albums/ab97/Crash_Nemesis/RED%20Dragon%20Tests/A023_C022_1114GH0004170_zps35432ef7.jpg (http://s853.photobucket.com/user/Crash_Nemesis/media/RED%20Dragon%20Tests/A023_C022_1114GH0004170_zps35432ef7.jpg.html)

2K
http://i853.photobucket.com/albums/ab97/Crash_Nemesis/RED%20Dragon%20Tests/A023_C025_1114H90001007_zps81002649.jpg (http://s853.photobucket.com/user/Crash_Nemesis/media/RED%20Dragon%20Tests/A023_C025_1114H90001007_zps81002649.jpg.html)

3K of my dog --

http://i853.photobucket.com/albums/ab97/Crash_Nemesis/RED%20Dragon%20Tests/A023_C017_1114W10000817_zps7e8b2cff.jpg~original

Larry Cheuk
11-17-2014, 03:05 PM
Thank you so much Christopher! Very helpful I really appreciate it!

Christopher-lee dos Santo
11-17-2014, 04:45 PM
Thank you so much Christopher! Very helpful I really appreciate it!

Anytime Larry. Glad I could help.