Official Low GOP topic, series 2

smsjr

Perry, I will take you up on the RED MX challenge. However, my talent will be a female. She's very quiet, and doesn't like lighting matches, so she'll instead be sitting by a single candle. I will use the ReAQuanted patch. For reference, what ISO did you use? I will also shoot this on the R1 MX. I will provide original files (mts and R3D) for download. I can create compressed versions as well, but I have to say that judging any of this stuff through Vimeo or Youtube is not very productive as it introduces too many variables. No processing to either file, other than developing the RED footage so that it looks normal in the RMD file, but anyone can download Redcine-X and do their own settings. I will do this not to say how much better R1 MX is, but rather to give everyone a baseline for where these patches are at and perhaps give some insight to how each camera is handling the image.

My thoughts on settings for GH2 would be the following:
Smooth Setting
Contrast: -2
Saturation: -1
Sharpness: -1
NR: -2

If anyone feels these settings would not be appropriate for any reason, please feel free to chime in before I do the test.

The shot it will attempt to simulate (not really, I am not that good and besides I'll be pulling my ow focus) can be found here: http://reduser.net/forum/showthread.php?40947-LEO-ISO-2000

Thoughts?

qwerty123

Why NR -2?

Your going to be at a high ISO, so NR should at least be at 0. See @cbrandin 's thread on Hack Myths and Realities.

Sharpness should be -2.

Make sure to white balance properly :-p (like I need to say that, but you'd be surprised at how often people fail to white balance appropriately).

driftwood

@kodakmoment I believe your problem lies with the SDXC 64GB. Even though its high spec card, the format of SDXC are proving worse than good old 30mb sandisk extremes or the 45Mb UHS-1. Even a decent Transcend type 10 has reported good returns. I think the bigger the card in some respects could also be the problem.
When I finally get my eval units in for a plethora of cards I will be reporting my results rest assured.
FYI Ive already binned a Panasonic 64Mb SDXC - the throughput just doesnt seem to add up.

kodakmoment

Thanks for your answer driftwood! I will give the other cards a try.

Ralph_B

@smsjr
This is a fascinating test - look forward to it. By all means keep NR at -2. GH2 noise reduction is EVIL, EVIL, EVIL! If you're shooting with a Panasonic lens, Qwerty is right, sharpness should be -2. However, if you're using a third party manual lens, the optimum sharpness could be anywhere from -2 to +2. Each lens is different and should be tested.

PerryWilson

@smsjr if you would like I can show you mine :P

smsjr

Reasoning on settings (I'm willing to shoot with a few different combos for comparison).

I use Nikon AIS primes. Not as sharp as those panasonic lenses, so don't feel the need to bring sharpness all the way down, but I can so you can compare. As for NR, I will try them all but I was thinking of turning it down all the way because RED doesn't use NR.

The lady will be wearing a grey shirt and might even hold a color chart for me.

Okay, on to the testing....

proaudio4

First of all, I really see no reason to compare with the RED MX. The RED MX is great and a better tool with MUCH better control.
But, if you want to do it for shits and giggles.... hehehe, why not. Of course, don't shoot at 4K and downscale to match. You are shooting 2k?

Second of all, I FULLY respect Chris and everything he has done here, simply amazing to say the least.

I have to agree with Ralph, and I've seen it myself, the GH2 NR is not the best.. that's for sure.
Remember there's only ONE knob, turn it up or down. This is not a serious tool for correct noise reduction.

And of course Chris is right if the NR is applied ahead of AVCHD compression, which it SHOULD be.
The concept is by using NR under noisy conditions, this will allow less bandwidth to be stolen for encoding noise.
But, the problem resides in the quality of NR reducing the overall image quality, so it's give and take.
Chris can still stand firm here since we need one-for-one comparisons with internal NR verses corrected NR in post. I have not done this test side-by-side, but just adjusting noise reduction in post is so much better due to the hack's higher bandwidth allowing finer natural noise. The post NR can make a better model for corrrection.

NR should be left in post with higher end solutions, pixel farm, Neat Video, etc...
Please leave it NR at -2 AND definately turn DOWN sharpness to -2 !
Just take a gander at the thread here on sharpness. Remember, sharpness does not add real detail. It enhances edge contrast creating horrible halos that make the image look fucking harsh.

f1lm3r

anyone else getting tired of reading about which SD cards work in a GOP thread? There's 4 "my SD card doesn't work" post to every 1 about the actual GOP patches. It's been covered so many times. If your card doesn't work, ask yourself, is it fast enough? where did you buy it? was it something from ebay? if it was, chances are you bought a fake. was it from an AUTHORIZED DEALER!?

smsjr

@perrywilson Once I finish and upload, let's compare. It will be interesting.

PerryWilson

sounds good lol

PerryWilson

man now I feel like I should reshoot my thing... want to replicate it

Ralph_B

@smsjr
While you're at it, why limit the test to a candle light. Why not shoot a normally lit scene, as well?

smsjr

@Ralph_B Already done that. This is just a fun little test to stress out the camera. Actually did one last night too under streetlight. Perry mentioned the Leo shot with MX and that seemed like a good low light test with GH2. This will be torture as I'll be doing a dolly move with very little light wide open on the lens. Always good to challenge yourself. :-)

strang3r1

my card is from amazon.com from an AUTHORIZED DEALER. Sandisk Extreme Pro 32 GB 45 MB/s card... I thought it was a really good one... still I'm doing everything correctly, I've tried a bunch of patches, none give me anything but choppy playback on the camera and computer... even got card read errors from some, crashing in really high light scenes. the most stable so far for me is the newest driftwood reaquainted, but still something doesn't seem right... my playback on my camera is SO choppy.

lolo

@smsjr thank you!! ,
but would you mind upload your test on this thread http://www.personal-view.com/talks/discussion/1030/gh2-versus-arri-alexa-and-f3-real-world-testing#Item_41 instead of this, so we can keep a track on every gh2 v/s high end camera test
please :)
and thank you again, it should be enlightening
:D

Ralph_B

@smsjr
Have you posted the earlier ones anywhere, or are they just an "in-house" project?

JDN

@strang3r1 and others with card problems, please note that you may have playback problems with these high bit rate patches on everything except the fastest computers -- I even get problems in VLC off a brand new macbook air from a SSD (never a problem in MplayerX though, so if you are on a mac, download that and your set). If you can, import it into any NLE first.

As for cards, my best experience has been with the Sandisk Extreme III cards (30 mb trasfer with a little "I" beside the SDHC logo). Also, an authorized dealer does not mean a legit supply chain -- that authorized dealer could have got duped. In North America, I only buy from Adorama, AbelCine, B&H, Henry's and Vistek -- I'm sure there are others that could be included, but you gotta pick a company with an established reputation. Paying $10 or $20 more is worth the piece of mind.

Sorry, to hear of your troubles though, hope it is just a playback issue.

smsjr

@Ralph_B Haven't posted those, and probably won't be able to. I will try to get some other stuff posted when possible. Trying to fit the testing in the small gaps in schedule.

strang3r1

@JDN thanks... the videos seem to playback fine if i upload them straight to vimeo... but I did actually order the SD card from Adorama and it still is giving me trouble... maybe I should try the 30 mb kind, but I don't understand why a faster version would pose issues. anyways - I probably need to upgrade my comp too, but it's the playback on the camera which is super choppy to the point of not playing back sometimes, which I don't understand.

Ralph_B

@smsjr
Thanks for your efforts!

smsjr

@lolo Yes, that thread does seem more appropriate. I will post there instead so that the focus can be specifically on the low GOP tests in this thread. Hopefully I didn't derail it too much.

disneytoy

Driftwood, do you have a 50 or 60 frame per second 720p GOP1 settings? I need to shoot some slow motion that can best cut with the latest (greatest) 1080p GOP1 settings. I wish we had 1080p 60. I'm sure it will be added in the GH-3. Sweet!

bkmcwd

From previous Q=16 version, this setting is changed so that the frame size under Lowlight may become large.
However, if it is used with 80% mode in high stress situation, recording may stop occasionally.
I think that this is only for 24p.
However, under lowlight, I also think, I-frame size about the same as Chris's 66M is saved.
Although my friend loved large frame size, I also made this for him. :-)

Q=16 version　is here
http://www.personal-view.com/talks/discussion/comment/21778#Comment_21778

BTW, when the benchmark test compared my own SanDisk 30M and 45M, Extream Pro 45M was a little faster than Extream 30M.

UPDATE:
The 1st patch is slightly unstable. 2nd patch is improved one.
The Streamparser chart of latest patch is here.
http://www.personal-view.com/talks/discussion/comment/22506#Comment_22506

driftwood

@bkmcwd Thats looking very good mate. And its stable? That DIY steadicam of yours is mighty fine. I want to buy one off you!

Howdy, Stranger!

Categories

Tags in Topic

Top Posters