Flaat settings for the NEX-5N

Samuel H

Samuel H

Major Contributor
DONE: Flaat for the 5N is out

I've tested it on my Flaat Shootout 2012, and I'm very, very happy with the results



My final settings are:
Flaat_10: Portrait, contrast=-1, saturation=-1, sharpness=-3, DR-1
Flaat_11: Portrait, contrast=-3, saturation=-1, sharpness=-3, DR-1

They're actually very similar to each other, and I'd rate them both at around 11 stops of DR, but Flaat_10 devotes less brightness values to the darkest stops of light, leaving more room for midtone detail.

Just like the previous ones, Flaat for the 5N is based on Portrait. But the reason here is different. In the case of Canon and Nikon, the reason is that I wanted to get the best skin tones possible. I also wanted that for the 5N, obviously, but, unlike on the Canon, I found that all picture styles on the 5N deliver relatively nice skin tones (i.e. none leaves me with the clay-looking people of Neutral on the Canons). the reason now is that Portrait has much wider DR than all the other options. No idea why, but that's what I've measured: about a stop more latitude with Portrait than with Standard or Sunset.


Flaat_10:
flaat_10.jpg


Flaat_11:
flaat_11.jpg
 
Last edited:
THIS POST HAS A LOG OF THE UPDATES WHICH LEAD TO THE STATE-OF-THE-ART, WHICH IS WHAT YOU CAN FIND IN THE FIRST POST



28-jun

For this one, I can't create a custom curve, like the ones I made for Canon and Nikon DSLRs. But I've been playing with the camera's settings, and so far I'm leaning towards some settings that I haven't read much about... so let's call it Flaat for the 5N
smiley.gif


First of all: once again, I'm using the Portrait picture style. On the Canons and Nikons it was because of the skin tones. On the 5N, skin tones seem good with all picture styles (none gives me those grey, clay-looking people from Canon's Neutral and its derivatives). But portrait has A LOT more DR than all the other options, so that's the one I'm using so far.

I still have to look closely into the downsides. There's a color shift in the reds, that's for sure, but I haven't tested these issues very throughly, yet (and it's not a major issue: it's the kind of thing you have to do to get better skin tones, Flaat for Canon definitely has this too). EOSHD reported that color gradients were smoother with sunset, but I have to pixel-peep before I can say how much worse Portrait is in this regard. Also, I'm a DR junkie, but I have to pixel-peep again to see if this DR that I'm getting is actually usable. I also have to calculate all the light response function, to make sure I'm not creating a wiggly response that is hard to grade. I won't be able to do all this until next week, but I'm excited about my findings, I'm loving this little camera so far, and I just couldn't keep my mouth shut any longer :p

My settings are:
Portrait, contrast=-3, saturation=-1, sharpness=-3

With this, I've measured some impressive DR numbers at ISO 100 (increasing this just to ISO 200 has a huge impact on DR!!):
(UPDATE: THESE MEASUREMENTS ARE FOR STILLS, IN VIDEO MODE THE NUMBERS CHANGE, A LOT)

DR-5
ISO 100: 12.0 stops (actually, slightly higher, but not quite 12.33 stops)
ISO 200: 10.5 stops
ISO 400: 10.5 stops
ISO 800: 10.5 stops
ISO 1600: 10.5 stops
ISO 3200: 10.5 stops

ISO 100:
DR-5: 12.00 stops (UPDATE: wiggly light response --> avoid!)
DR-4: 11.33 stops
DR-3: 11.00 stops (this would be Flaat_11p)
DR-2: 10.66 stops
DR-1: 10.33 stops
DR-0: 10.00 stops (this would be Flaat_10p)

ISO 800:
DR-5: 10.5 stops
DR-4: 10 stops
DR-3: 10 stops
DR-2: 10 stops
DR-1: 10 stops
DR-0: 9.66 stops


30-jun

I originally measured DR in stills mode, and it turns out that, unlike what I've seen in the Canons, the in-camera processing of stills and video leads to different results, probably because of the different color space; I still see better DR with Portrait than with other creative styles, but it's not THAT good.

I've also checked that the widest setting leads to a wiggly light response, so I would advise against DR-5 until I can check whether there's a way to avoid this


2-jul

I have uploaded a bunch of measured light-response functions, here:
http://www.similaar.com/foto/flaat-5n/technical-tests-1.html

I still have to test the more conservative options (e.g. contrast=-2 with DR-3), but right now my Flaat settings would be:
Portrait, contrast=-3, saturation=-2, sharpness=-3, DR-1
 
Last edited:
Love the NEX5N as well. Will check out your settings once I'm back home in a few weeks. I've been using the Sunset profile and the DR isn't great in that, but I'm also rarely at ISO 100, so maybe that's the problem.
 
I've managed to pixel-peep just a bit. I'm not in my place, so I can't use my monitor, I have no editing software, etc. But... so far it's looking great!!

* I have seen no gradient or macroblocking issues. This may change when I load the footage in the editor and try to push it.

* I like the colors. The shift in the reds and oranges is quite obvious (Portrait seems to push everything around "skin" towards "skin"), but I can live with that. Specially if skin tones are nice, which seems to be the case (no vectorscope, and uncalibrated monitor, so: confirmation pending).

* I like the sharpness/softness of the image. It's not clinically sharp, as with the GH2, but it's not visibly soft either, as it would be on the Canons when shooting with Sharpness=0. For most of my footage, I don't think I'll add sharpness in post.

* Very positive surprise: aliasing/moire issues are greatly reduced with my settings. Aliasing is only an issue for very extreme cases, like near-horizontal cables in front of a bright sky, and even then it's much less obvious than with, say, Sunset (even with everything else just the same as in my settings, including sharpness=-3). Lack of moire is even more surprising: I've tried to get moire issues by shooting 4 different problematic targets, and 3 of them came out 100% clean: I found no issues whatsoever on my short clips of floor-made-of-little-stones, brick buildings, and tiled roofs, all of them shot in just the way that would have made my Canon cry like a baby. I'm sure I could make the moire patterns appear, because the fourth test (problematic cloth pattern) showed it, but it's definitely an accident (I shot it at 3 different distances, and only one shows moire), not the norm, as it is with the old Canons.
 
Last edited:
I am yet to critically examine the difference between dr optimisation on or off. I use sunset mode myself. something nice about the graduation between light and dark using this mode.

i also in two minds about the benefits of shooting flat on the nex5n. at the moment i am trying to get as close as possible to the final result in camera with minor tweeks. instead of shooting flat and applying more contrast and colour adjustment. cheers for the dr test. i may well try the portrait mode again.
 
One thing that REALLY bothers me about the Sony 5n "dynamic range" is that all the shadows are noisy in a blotchy way. Shooting at ISO 1600 with a low contrast brings up some really bad darks. Way worse than my 7d or 5d mark III. That's why I decided to lot DR be and settled for the -1 contrast setting instead of -3. That DR down there is unusable to me.

At ISO 100 it looks kinda ok but now the -3 contrast shadows are just really grey. No saturation there at all.
 
I've shot some footage with my widest settings (Flaat_12p?) at ISO 3200, and the blacks are indeed awful.
I still have to treat that footage with MB Denoiser to see what's useful and what has to be crushed.

In any case, it's good that it's me (and not the camera) who decides what gets crushed and what doesn't. And this problem doesn't appear at ISO 100.
 
Samuel H said:
I've shot some footage with my widest settings (Flaat_12p?) at ISO 3200, and the blacks are indeed awful.

And this problem doesn't appear at ISO 100.
Click to expand...

I have to disagree. At ISO 100 I still see ugly shadow noise when using Portrait -3, -1, -3 and DRO 5. Portrait always seems to give noisy shadows so you're not really getting any extra detail by having those blacks lifted up so far.

I prefer Sunset -3, -3, -3 and only using DRO 5 at ISO 100. All other ISO settings I leave DRO at 0. At ISO 1600 I like to put the Contrast to -2 and at ISO 3200 Contrast to -1. It's also important at these higher ISO settings to overexpose the image to hide the shadow noise. At ISO 3200 if you can get everything in the image above about 25 IRE there's almost no visible noise.

In my testing I found the way that midtone contrast was rendered in Sunset to be slightly better than using the Standard picture profile at -3, -1, -3, but it wasn't a huge difference.
 
I'm not sure about that, but I have found some hints that DR-5 may be a bad idea: the light response becomes wiggly, with a bump at around IRE 30%. This is highly undesirable, because, if confirmed, it will lead to unpredictable results and difficult-to-grade footage. So far I'm back to DR-3, which doesn't seem to have this problem.

Also, my DR measurements above were in stills mode. I always do it that way, because I found that with the Canons the results were exactly the same, but it's much easier to work with stills. The 5N leads to different results in stills and video mode, probably because the color space is different, so the measurements above are not accurate. I still see more DR in portrait than in sunset, but I'll have to measure it differently to see how big the difference is, and if this camera is actually giving me more DR than my old Canon (I'll run side-by-side tests next week).
 
I've built an infrastructure for calculating the light-response functions directly in video mode, without relying on stills. I can confirm that DR is nearly a stop smaller for video than for jpg stills, and that my original settings with DR-5 deliver a nasty, wiggly light response. I'll try to calculate the light response function with different settings in the following couple of days, and see if I can come out with something that resembles what Flaat does for Canon and Nikon DSLRs. I still like the colors and "softness" of portrait better than what I get from sunset, and DR seems wider, but the big prize is for a near-log light response across 11 stops of latitude :)
 
After measuring a few light response functions (you can see them here: http://www.similaar.com/foto/flaat-5n/technical-tests-1.html), now my favored settings are:
Portrait, contrast=-3, saturation=-1, sharpness=-3, DR-1

With this, in video mode, I've measured 11 stops of latitude at ISO 100, with a light response function that is, well, not too bad; I would call it "piecewise log"; not what I'd settle for in the Canons or Nikons, but not a bad option if I can't define my own curve, and in any case the most "neutral" one I've been able to find yet (moderately wide, not wiggly, and with as little of an in-built s-curve as I could get)
p_c3_dr1.jpg



(I updated the first two posts of the thread to reflect my recent findings)
 
Last edited:
You must log in or register to reply here.
Back
Top