FS700 AVCHD vs Odyssey 7Q 2K DNG

McReedy

McReedy

Active member
I did a small test to compare the internal AVCHD vs the Odyssey 7Q 2K DNG and got some impressive results.

The O7Q not only keeps the same 2K DNG sharpness from 24, 120 and 240fps it also shows a much higher detail compared to the "digital" looking internal AVCHD (no sharpness applied).

Here you can download the frames (16MB):
https://dl.dropboxusercontent.com/u/3545751/fs700-compare/fs700_avchd_vs_o7q_frames.zip

Specially on the moving picture the difference is visible:

And here's the 2048x1080 ProRes 422(HQ) result (450MB):
https://dl.dropboxusercontent.com/u/3545751/fs700-compare/fs700_avchd_vs_o7q.mov.zip

The amount of detail the O7Q delivers is pretty the limit of Full HD!
 

Attachments

  • fs700_avchd_vs_o7q_allframes.jpg
    fs700_avchd_vs_o7q_allframes.jpg
    19.8 KB · Views: 0
  • fs700_avchd_vs_o7q_240fps.jpg
    fs700_avchd_vs_o7q_240fps.jpg
    20.6 KB · Views: 0
Last edited:
Thanks for posting. The difference between AVCHD and DNG is quite noticeable. I'm glad the 240 fps video looks so good.
 
The 24 fps is visually better than 240 fps in stock mode. I'd be curious to see 1080 at 24 vs odyssey 2k at 24, If you can post samples that it would be great. Thanks for the post
 
Thanks for posting these frames - really interesting to see.

Marino, the OP actually posted sample frames from several frame rates from 2K RAW and AVCHD, if you download the zip file. Check em out - it's quite interesting to see the difference.
 
Thanks for the comparison McReedy! Love the lattitude and detail in those DNG's you posted. Matthew Allard did a review of 240fps raw over on newshooter and said he was pretty frustrated by the noise in slog2 at high frame rates. I can definitely see a rise in noise in the 240fps footage, but I'm guessing that's just a trade off of the required higher shutter speed causing the bump in ISO and seems pretty manageable to me with a little noise reduction in post. I appreciate you posting your findings while we wait for our pre-orders to role out.

Hurry up and get here 7q! Gah!
 
Slog2 suffers from a fair amount of noise in the shadows. This is just the tradeoff you get with Slog2; and, really is not related to the ISO. When using Slog2, one needs to be exceptionally careful with the exposure, taking care never to underexpose. If exposure error is expected, you're much better over-exposing and bringing things down in post, including noise signature in the shadows. ETTR is always recommended, altho, you never want to blow out the highlites. 18% gray with Slog2, is around IRE 32-40.
 
Last edited:
brazzam said:
Thanks for posting these frames - really interesting to see.

Marino, the OP actually posted sample frames from several frame rates from 2K RAW and AVCHD, if you download the zip file. Check em out - it's quite interesting to see the difference.
Click to expand...

Thanks, the issues with my iPad, can't download from here. I'll check them out on my desktop.
 
There is one thing that i cannot understand.
If everything is constant (ISO, shutter, apperture, picture style, etc), why there is a significant noise increase on the 240fps RAW frame when compared with the 120fps or 24fps? After all, isnt it raw?
Shouldn't it be exactly the same frame at whatever frame rate?
 
Let's imagine that the settings from the frames on the first post are:
1/250, f/4.0, ISO 2000.
You can shoot at whatever frame rate you want and the shutter and aperture can be always the same.

I suppose that the OP did this, set the same settings at every frame rate (otherwise the comparison wouldn't make any sense), so, there is the question again, why more noise in 240fps? I find it hard to understand.

Could anyone explain this to me?
 
Just a mildly educated guess: Pulling the information off of the CMOS 240 times per second is significantly more processor intensive than doing so 24 times per second. The more processor intensive something is, the more heat is generated. Greater heat=greater resistance. Greater resistance can lead to more noise. Could be other factors like slightly more abbreviated image processing algorithms, but I would say that pulling info off the CMOS more often would be the reason for greater noise.
 
As hugofilipe wrote I used the same shutterspeed to get the same image in all frame rates, because nd, iso or aperture would change the image itself. The Shot was ETTR oriented on the bright cloudy sky. With Highlightrecover you really can get more detail in the very bright areas - its really impressive!
If you record the buffered 240fps avchd it uses the same 2K Rawstream - you can see the same wavelet compressionartifacts on some spots in the image - but the post-debayering brings us much more detail. The new Davinci 10.1 handles the DNG Frames much faster now btw!
 
I inicially though of that, could be heat generated that was causing the noise, I've seen it on other cameras.
I cannot test my self since my ordered Odyssey 7Q did not arrive yet, but I'm pretty sure that if we compare de first frame of a 24fps footage and the first of a 240fps, we'll see more noise on the 240 than on the 24, and we can't blame the heat.

I have to be honest, I'm a bit disampointed with the FS700 and all my hope was that RAW or a better codec than the AVCHD could "save it" but my hopes are going away. (sentimental part :) ) ISO noise is incredibly high at 240fps, even in RAW footage. The reviews I read prior to purschase was that it was a "low light monster", "no noise at high iso's", etc. I have the speedbooster and I use really wide aperture lenses when needed (mainly f1.4 (1.0 on SB)) and i really never leave the ISO 2000. I feel sad with it that i cannot even imagine using higher ISO's.

But I still have hope!

McReedy, could you share some RAW footage? Something with lots of highlights and shadows, maybe a direct sunset with shadow areas in front, anything you find usefull to see the camera true lattitude.

Thank you guys!
 
hugofilipe,

I can understand your initial disappointment with the noise levels on the FS700. The SLog2 profile seems to be inherently noisy when exposed normally. It really benefits from ETTR, that is, exposing so that the highlights are practically blown out. Then, in post, you have to bring the exposure down. That really eliminates the noise. Once you get your O7Q, things will be a little better, but, not by a lot. You have to remember that the Slog2 gamma curve pushes all the info into the upper end of the gamma curve, with very few bits left for the shadows. Once you record RAW with the O7Q, I've found the best exposure to be in the range where an 18% gray card is exposed for 32-40% IRE.
 
If you treat the Odyssey WF as if it is reading 20 points to high in 2Kraw and set your exposure accordingly, it works great, low noise great highlight recovery. Set skin tones to 50-55, not 38.

This will be fixed in future FW. Also white point is off in the current FW but this will also be fixed.

There have been numerous posts her in the last few months with the same conclusions.

Right now it is a great monitor, I would say the best field monitor I ever used, and 240 FPS 2K recorder, and soon it will be a lot more. It is great value with current firmware you just have to understand the limitations of living on the bleeding edge. With new 4K ProRes etc it will be the only thing in its class. It is a win win.
 
olof said:
If you treat the Odyssey WF as if it is reading 20 points to high in 2Kraw and set your exposure accordingly, it works great, low noise great highlight recovery. Set skin tones to 50-55, not 38.

This will be fixed in future FW. Also white point is off in the current FW but this will also be fixed.
Click to expand...

Olof,
Just FYI....I have my artificial color settings so that green displays 32-40% This is a pretty accurate way of setting exposure. When the green color gives way to light gray, I know it's 40%. When the green gives way to dark gray, I know it's 32%. When there's no gray card to expose to, I can make an informed value judgment as to what's a good exposure. I've been using this method for a few weeks now, and it hasn't let me down.



As you said, skintones work best around 50%, but, for a landscape photographer, skintones aren't always in frame. But, with any expereince, it's pretty easy to know what approximates 18% gray.

P.S. I just got your O7Q sunshade. I love it! Works perfectly, especially in very bright sunlight.
 
Last edited:
olof said:
If you treat the Odyssey WF as if it is reading 20 points to high in 2Kraw and set your exposure accordingly, it works great, low noise great highlight recovery. Set skin tones to 50-55, not 38.

This will be fixed in future FW. Also white point is off in the current FW but this will also be fixed.
Click to expand...

Hi Olof, you are much more technical than myself. So what does it mean exactly when you say the "White point is off?"

Also, can you explain to me what you mean with "Set skin tones to 50-55, not 38."

Thanks so much.

At HugoFlippa, the FS700 with 7Q is much better than AVCHD for sure. Here's some 7Q test shots I did that have some real lattitude (sunsets), that would toast AVCHD. I really liked what I was getting from the 7Q with these shots.

http://vimeo.com/83642889
 
gene_can_sing said:
At HugoFlippa, the FS700 with 7Q is much better than AVCHD for sure. Here's some 7Q test shots I did that have some real lattitude (sunsets), that would toast AVCHD. I really liked what I was getting from the 7Q with these shots.

http://vimeo.com/83642889
Click to expand...

Thanks for sharing.
At 00:37 I would expect that was possible to take a bit more of highlight detail from that window or even on the last few shots, more cloud details on the sky. Shouldn't the sensor provide 13-14fstops?
I feel that a 11fstop DR camera could do better.

I'm just being impacient, I have to wait for the shipment of my Odyssey do see it for my self. I hope that I'll be impressed with it.

Can anyone provide a CinemaDNG clip to play with? via Dropbox folder or Wetransfer. I tried to download the Convergent Design ProRes examples but looks like its offline.

Thanks!
 
hugofilipe said:
Thanks for sharing.
At 00:37 I would expect that was possible to take a bit more of highlight detail from that window or even on the last few shots, more cloud details on the sky. Shouldn't the sensor provide 13-14fstops?
I feel that a 11fstop DR camera could do better.

I'm just being impacient, I have to wait for the shipment of my Odyssey do see it for my self. I hope that I'll be impressed with it.

Can anyone provide a CinemaDNG clip to play with? via Dropbox folder or Wetransfer. I tried to download the Convergent Design ProRes examples but looks like its offline.

Thanks!
Click to expand...

Maybe your expectations are too high. As someone who shot tons with 11 stop cameras (Canon DSLRs), I can assure you that S-Log is in another league. Maybe it's best to just pony up the money for an F5, a better camera for sure. I believe a decent F5 package can be had for a little under $25K with 4K recorder. I guess if you want it all, you gotta just pay the price. For myself, I can't justify that price yet, but yeah... I sure would love an F5.
 
You must log in or register to reply here.
Back
Top