Awesome FS700 test of the FS700

[FelixGER]

Edit: Just corrected the epic fail of naming the topic, but it doesn´t work, lol.

Adam Wilt wrote a very detailed article:

http://provideocoalition.com/index....ony_nex-fs700_super35_lss_avchd_camcorder/P0/

Sony says the FS700 uses “approximately 8,300,000 pixels” in creating its 16x9 image, which is close enough (the sensor has about 11,600,000 pixels or photosites overall). This high pixel count means that the camera should have more than enough oversampling to render a very crisp, sharp HD image at the limits of what 1920x1080 can show: up to 1080 TVl/ph in both H and V directions.

Surprisingly, the FS700 is very close to the FS100 on the charts, capturing 800 TVl/ph horizontally and 1000+ lines vertically [...]

Mehhhh! So Frank, you were a little too enthusiastic when putting it next to the C300
I ordered it anyway and an Atomos Samurai...If the C300 was just a little cheaper...

 

[paulcurtis]

Mehhhh! So Frank, you were a little too enthusiastic when putting it next to the C300
I ordered it anyway and an Atomos Samurai...If the C300 was just a little cheaper...

A lot of resolution loss is going to happen during compression, so it really does depend *how* you're recording or measuring the resolution and whether you have any realistic way of capturing it.

I saw an excellent comparison of ArriRAW and ProRes 444. The difference between the two was very, very obvious and the ProRes showed substantially less resolution. And that's the high end version of ProRes too.

So we have AVCHD at worst, 422 ProRes and uncompressed 422 at best in our portable recorders. Well that's decimating the signal quite a bit. We must remember that ProRes, Cineform, et al are not lossless. Stuff is thrown away. Resolution, especially low contrast resolution is the first to go.

What i'd be interested in doing would be sticking the out of the 700 into a 444 capture card and seeing how that looks (just need the 700). Sort of immaterial because you couldn't do that on the run but it might show what the sensor is capable of at best. Your resolution is only as good as the worst link, be that lens, sensor or recorder.

Anyway, really resolution is so overrated most of the time. Where's that 12bit 4k announcement....

An awful lot of the C300 res seems like sharpened artefacts to my eye (from the few samples i've seen)

cheers
paul

 

[FelixGER]

The thing with the Alexa might be that Prores is scaled to 1080p while ArriRaw is 2880 x 2160 (2,9k)

Sorry, but an EX1 with its 35mbit mpeg2 and e GH2 with 24 mbit (a lot worse than the FS100/700s) AVCHD manages to get 1000 lines of resolution. If you needed RAW files to "see" Full HD what use would a movie on bluray disc be?

I´m looking forward for the RAW update but I´m not expecting 4k but real 1080p. That´s enough for me but the minimum, too.

Dynamic Range test:
http://provideocoalition.com/index....ras_sony_fs700_dynamic_range_presentation/P1/

 

[paulcurtis]

Without really trying for myself, from what i've seen the 700 looks more detailed than the 100, so that's an odd conclusion. Do you know whether there are any tests with production versions of the camera?

I noticed a C300 vs FS100 resolution test on vimeo the other day and it just looked like the 100 was out of focus. The lens and aperture can make such a huge difference that unless they're careful matched (ideally the same lens) the results are moot.

I'm not saying the results are wrong, just find them a little odd really. Especially as we're doing a 4k downsample (it's certainly not line skipping that's for sure). I can't remember but i believe it was 1080p vs scaled ArriRaw so the end result was 1080. It was just revealing how much softer the Prores stuff was - it surprised me and just got me thinking about other situations where we're taking a hit

cheers
paul

 

[vincent thomas]

I wish the c300 has better highlight roll off with higher dynamic range, but for the rest, i really don't see what you are talking, the image feels fantastic

vince

A lot of resolution loss is going to happen during compression, so it really does depend *how* you're recording or measuring the resolution and whether you have any realistic way of capturing it.



An awful lot of the C300 res seems like sharpened artefacts to my eye (from the few samples i've seen)

cheers
paul

 

[Postmaster]

E
Mehhhh! So Frank, you were a little too enthusiastic when putting it next to the C300
I ordered it anyway and an Atomos Samurai...If the C300 was just a little cheaper...

Nah, I wasn't.

When you point both camera to real world scenarios, instead of black & white charts, the 700 is -at lest - in the same league as the C300 , if not better.
You can even see that in Philip Blooms resolution test with the FS100 vs C300.

Frank

 

[paulcurtis]

I wish the c300 has better highlight roll off with higher dynamic range, but for the rest, i really don't see what you are talking, the image feels fantastic

vince

I'm certainly not bashing the C300, it is a wonderful image overall but all i'm saying is whenever i've seen full res stills from frames they've had that sharpened look and seem quite susceptible to aliasing type noise (like the FS in highlight situations). I find the fact that it's not totally alias free surprising as it's scaled down.

cheers
paul

 

[rotierney]

Just uploaded my first impressions and my first test short here:

http://www.rotierney.co.nz/?p=1568

 

[Razz16mm]

If you want a smooth film like non-aliased image, 800-850 TVL vertical and about 1.4k -1.6k horizontal is as good as it gets for 1080p.

 

[Gene Sung]

Hey Ro, Your footage looked really, really good; both in shooting and treatment. One of the much better FS700 tests for sure in terms of a cinematic look. Lots of the FS700 videos I've been have this really clean, saturated, plastic look which you entirely avoided, which makes me really happy.

You make me want to call the Rep that I ordered the camera from, but I just got done bothering him several days ago. I want my FS700 now! LOL.

Can I ask you a question? You mentioned that you modified your Cine Gamma 1 profile a bit. What did you do to your profile? I love the muted, less saturated look that you got out of the camera. Your image just has so much character compared to a lot of the plasticy images that I have seen.

 

[Jay Birch]

Very nice, rotierney... Some good DR on show. I find it a little hard to swallow the 14stops the 1st review mentions (maybe when 12bit RAW is added)... But this camera seems to be getting towards 12 stops, which I personally consider the magic number.

 

[rotierney]

Hey Ro, Your footage looked really, really good; both in shooting and treatment. One of the much better FS700 tests for sure in terms of a cinematic look. Lots of the FS700 videos I've been have this really clean, saturated, plastic look which you entirely avoided, which makes me really happy.

You make me want to call the Rep that I ordered the camera from, but I just got done bothering him several days ago. I want my FS700 now! LOL.

Can I ask you a question? You mentioned that you modified your Cine Gamma 1 profile a bit. What did you do to your profile? I love the muted, less saturated look that you got out of the camera. Your image just has so much character compared to a lot of the plasticy images that I have seen.

Thanks for your feedback. Cine gamma tweaks were minor adjustments to the detail settings and colour, but mainly raising the black level up to +5 to create a film wash look. I then added a little contrast and warmth in post.

 

[rotierney]

Very nice, rotierney... Some good DR on show. I find it a little hard to swallow the 14stops the 1st review mentions (maybe when 12bit RAW is added)... But this camera seems to be getting towards 12 stops, which I personally consider the magic number.

Yes I am blown away with the DR on this camera. I haven't done any scientific tests but is seems more than enough for real life shooting.

 

[Cory Braun]

Very nice, rotierney... Some good DR on show. I find it a little hard to swallow the 14stops the 1st review mentions (maybe when 12bit RAW is added)... But this camera seems to be getting towards 12 stops, which I personally consider the magic number.

He mentions throughout the article multiple times that it has 14 stops of DR, and then once or twice throws in there that it's only 12 usable stops. Kind of annoying. Now other sites are using the article as a source and it's spreading that it has 14 stops. I believe the Alexa is 14 stops... from the clips I have seen, this is not reaching Alexa territory.

 

[paulcurtis]

Thanks for your feedback. Cine gamma tweaks were minor adjustments to the detail settings and colour, but mainly raising the black level up to +5 to create a film wash look. I then added a little contrast and warmth in post.

Without looking carefully at the source files (especially with the new gammas) you stand a good chance of missing out on some range by doing that. It does look like your files are not going all the way to black although you may have graded them that way after. There's little point in pushing up your blacks in camera, instead shift everything down, get a little more highlight range and then push the blacks up after if that's what you want.

The 14 stops comment is about the sensor itself, of course to see 14 separate stops squeezed into an 8 bit container then your lowest or highest stops are going to have so little separation as to be irrelevent. However should the promise of RAW arrive then perhaps you will actually get 14 stops of useable range - impossible to say until we actually see it, but the idea is that the sensor could be capable of distinguishing that kind of range.

cheers
paul

 

[rotierney]

There's little point in pushing up your blacks in camera, instead shift everything down, get a little more highlight range and then push the blacks up after if that's what you want.

Thanks Paul. That is very useful advice. I guess I'm just too used to pushing blacks up on Canon H264 footage in post and getting that horrific blotchy noise. Still very much in experimentation mode with this camera.

 

[Gene Sung]

Hey Rotierney, did you turn the Detail setting down or up? I'm guessing you turned it down a bit since it looks very cinematic. How many notches did you tweak the detail setting?

Great work!

 

[rotierney]

Hey Rotierney, did you turn the Detail setting down or up? I'm guessing you turned it down a bit since it looks very cinematic. How many notches did you tweak the detail setting?

Great work!

I dialed the detail setting down to -3 when I used the Sigma 50mm F1.4 lens. This was so that it matched the softer old Nikkor 50mm.

 

[paulcurtis]

Thanks Paul. That is very useful advice. I guess I'm just too used to pushing blacks up on Canon H264 footage in post and getting that horrific blotchy noise. Still very much in experimentation mode with this camera.

There are a few threads on here about this, i hope that would be very useful. Nirv experimented with black levels a bit and that's documented. The camera is really super clean in the shadows and by default there's a tendency for the 100 to not go all the way to black. It seems that the output can be the broadcast safe 16-235. We all know about superbrights in the highlights but there are also superblacks below 16 and sometimes you even need a negative black level to get there.

The 700 may well be different though, i'm looking forward to getting mine.

Your footage shows some good transitions to over exposure.

I assume you don't have a 100 as well, so you can't compare sensitivity? Some say it's pretty much the same others say less sensitive.

cheers
Paul

 

[Jay Birch]

from the clips I have seen, this is not reaching Alexa territory.

Nowhere close... It seems like people can't trust their own eyes any more, just tech charts (and charts can be misinterpreted). My eyes tell me it has around 11-12 stops when really flat. Would be interesting to here from the more technical users on here what they think, i'd loved to be proven wrong and that these guys have dialled in the settings to get 14 stops.