Sony EX 3 filmout

[Bobby Shore]

hey,

 

So I've just started prep on a feature that'll be shooting on the Sony EX 3 and finishing to film for theatrical distribution. I've had a chance to test out the workflow, albeit in a pretty basic way (was comparing different camera's, RED, EX 3, 5D and 7D). I'm trying to get production on board for something a little more comprehensive focusing only on the EX 3 - mostly to test detail settings in camera, gamma, etc. in order to strike the best print.

 

The feature is a follow up to a canadian moc doc shot 8 years ago called Fubar - was ultra low budget, had a really lo-fi feel to it (was shot on an Xl 1). This time around there's a good budget, lot of production support, etc., but the aesthetic has to remain in the same ballpark as the first. It's pseudo moc doc again (not strictly adhering to those rules for the whole shoot), so the aesthetic will run along the lines as the first - handheld, lo-fi, available light type look, etc. I'm working really closely with the production designer to get as much practical lighting into the sets and locations as possible. There's no blocking, the entire script is improvised, and we have to be prepped to shoot 360 almost always.

 

I'm posting cause I'm trying to gather as much info about filming out from digital formats as possible. I was able to test a couple settings with the EX 3 - ie. pushing the gain, lattitude, etc., but if anyone has experience with the process, I'd love to hear what you have to say. Like I said, I'm trying to shoot more tests, but the more I know before the more specific I can get with the testing. I've found some info on the boards from David Williams that was really helpful:

 

Detail is really subject to personal taste. I have shot a few low budget features, and have seen some 35mm blow up tests done that looked quite good using these detail settings-

 

level -20

frequency +45

Crispening +20

White Limit +80

Black Limit +80

Knee Apt +10

 

Using CineGamma 1 mostly, and Matrix set to Cinema, Level +30. That shoots flat, great for post grading.

 

Change as required of course, depending on conditions and scene. Avoid CineGamma 2 as it limits range to 100%, the other are 109%. 1 stretches highlights, 3 is average, 4 stretches blacks.

 

If anyone has any other advice, info, opinions, I'd love to hear them. Thanks in advance.

 

Bobby Shore

mtl/la

www.bobbyshore.com

[Pascal Lebegue]

Thank you Bobbe, I am myself trying to pick the best option for a movie project, out of the hi-end HD cameras...

and your crisp and straight forward informations are needed.

[Bobby Shore]

Thank you Bobbe, I am myself trying to pick the best option for a movie project, out of the hi-end HD cameras...

and your crisp and straight forward informations are needed.

 

I'm heading out today to shoot something a little more comprehensive in terms of striking a print from the EX 3: testing detail settings (on, off, and a couple variations like the settings posted above), as well as some different gamma options. I was hoping to shoot this round of tests with the Convergent Design nanoflash recorder, which allows the EX to capture 4:2:2 instead of 4:2:0, but unfortunately it's not available. I'll be using one once principal starts though.

 

Also, as per an earlier post regarding resolving power of the EX 3 vs the canon 7D here, I'm also gonna shoot a side by side night ext. WS (available light), and see how the two render next to eachother on the print. This shot will be static, so as to avoid any of the movement issues the 7D has.

 

I'll post results when I check the print next week. In the meantime, any advice, opinions, experiences, etc., would be more than welcome. Thanks.

 

bobby

[Miguel Bunster]

Bobby,

I am using the nanoflahs on the f900r but keep in mind to get the fast cards like the ultra IV (i am using extreme III) and it can only get now a 140mbs data rate (which is pretty much the data rate for the hdcam tape on 3:1:1). IF you get the faster cards you can go up to 220mbs on 4:2:2

best

m

[Bobby Shore]

Bobby,

I am using the nanoflahs on the f900r but keep in mind to get the fast cards like the ultra IV (i am using extreme III) and it can only get now a 140mbs data rate (which is pretty much the data rate for the hdcam tape on 3:1:1). IF you get the faster cards you can go up to 220mbs on 4:2:2

best

m

 

yo man!! how you doing? The sweet spot for the EX and nanoflash (from the research I've done), is floating around 100mbs (while still shooting 4:2:2), anymore and the added image quality is actually quite minimal in respect to the added memory required for post work. Any advice for using the thing? Issues at all? I'm testing it in a couple days, but if you have a heads up on inside info on it, I'd love to hear it.

 

hope the feature's going well man. keep in touch. lates!

 

bobby

[Miguel Bunster]

Hey man,

I really like the nanoflash, is small and practical and you get alot out of the 32GB cards. Only a couple times I would have problems like i press record and it wouldn't even if the card wasn't full but then switched cards and no problem. Nice to playback shots and so on.

 

Yes I read the same that the sweet spot is 100mbs and called the company and the same info. I was curios that HDcam is putting 1440x1080 with a 3:1:1 on a 144mbs and the nano flash is putting more info in the same bandwidth which is 1920x1080 at 4:2:2 so there is more compression I guess but was explained its still better than the other option. If data management is not a problem why not get a better bit rate?

 

I was told that is better to shoot Long Gop compression than I-frame on the nano flash as well.

 

the other thing to check is that if the HDSI of the EX3 outputs sound as well. It should. The F900R does too.

Make sure you are not outputting through the HDSI (it shouldnt) any on screen info like characters, menu etc because the nanoflash will record "Anything" you give it. As well I think you can use it with the 7d because it has an HDMI input.

 

Movie is going well man!

you?

t

best

m

[Bobby Shore]

so I was able to check out the 35mm print from the camera tests last week, was mostly checking out the effects of shooting with or without detail on the EX 3 and how it'd translate to the print. Also, did a side by side with the 7D.

 

with the detail, I tested it both on, off, and then with a custom setting (ref. above) dialed in. I found the print looked best with the detail set ON. It added a certain crispness and sharpness to the image, but it didn't at all look like false edge enhancement. with the detail set OFF, the overall quality was kinda mushy. I guess I'd compare the discrepancy almost like using two different lenses, one being sharper and appearing to resolve more than the other. The custom setting was still a little on the softer side for my taste. I was pretty amazed at how smooth the crispness of the image appeared with the detail on... I'm not sure if that comes from there being a slight generation loss going to film (was an arri laser, so apparently there is no gen loss), or if it was just the analog look of the film taking the curse off the digital edge enhancement. Either way, definitely looked best to my eye.

 

only problem now is that for the print, I'd defintiely keep the detail on, but for the HD finish I'd prefer to shoot with the detail OFF (just personal preference, I feel like detail setting when finishing HD looks too crisp and enhanced). I know that you can diminish the effective detail in post, but the post place is saying it'd be too time consuming/expensive. Now I'm trying to decide on shooting for the print or shooting for the non-theatrical finish. Or maybe just split the difference.

 

For the other tests, night ext. side by side, the 7D won out hands down. I was under a sodium vap. street light in a parking lot, which was reading 1.4 rated @ 500 asa (which I've been rating the EX 3 at). The asphalt on the ground was spotting @ 0.7 6/10, and white car under the light was spotting @ 2.0 1/3) The EX held up, shooting wide open at plus 3 db, but the image was still a touch dark on the print (no color correct or anything was done). I could've even pushed the gain to 6 and the noise would've been minimal. However, the 7D (which was rated at 1000 with an L series 24-70 zoom shot at a 2.8) gave me the same amount of info in the frame (in terms of overall luminosity) even though it was shooting one stop down from the EX. I think with a fast prime lens, the 7D would work great for some of the night ext. establishing shots we'll be getting for the show that we won't have the firepower to light up to a decent stop. Also, even at 1000 ASA and projected huge, the image was still very clean looking, could definitely push it to 1600 easy. And the footage between the two camera's will be able to intercut for sure, as long as the 7D is kept to the wide side of the lens and with minimal camera movement (no jello cam).

 

Definitely an interesting test. Any thoughts, comments, etc., are more than welcome. thanks.

 

Bobby Shore

DP

mtl/la

www.bobbyshore.com

[Martin Hawkes]

Definitely an interesting test. Any thoughts, comments, etc., are more than welcome. thanks.

 

Bobby Shore

DP

mtl/la

www.bobbyshore.com

 

 

Hi Bobby,

 

Any chance of you posting stills from the side by side test?

 

Sounds like a nice gig. Now you'll just have to giver!

 

All the best,

 

Martin

[Bobby Shore]

Hi Bobby,

 

Any chance of you posting stills from the side by side test?

 

Sounds like a nice gig. Now you'll just have to giver!

 

All the best,

 

Martin

 

here's a link to a rough cut of the test the editor laid out (here). It's pretty low res, so it's hard to really notice any distinctive differences. The black frames separating shots were removed to get a better sense of changes shot to shot, and at the end there's three shots of EX night ext. before the 7D footy. Only pay attention to the last EX 3 shot (is where I bumped the gain to + 3 to better match the 7D - which with a fast prime could give me two more stops than what is shown in the test, which is why I'm planning on using the 7D for wide night establishing shots for the show).

 

for the detail and gamma tests. the key was a tungsten based diva light direct, reading 4.0 @ 500 asa. fill side was 1.4. BG was practical cool whites (green bias), spotting @ 2.8 8/10. The chart on the frame right was spotting @ 11.0 1/10 on the whites, and @ 1.0 on the blacks (including the surrounding black areas).

 

The art dept. just bought a palette of pilsner (4000 cans), a third of which they want emptied for set dressing... guess it's one of the perks of making a movie about drunken banger's.

 

bobby

[Bill Totolo]

here's a link to a rough cut of the test the editor laid out (here).

The video is set to "private".

[Bobby Shore]

The video is set to "private".

 

sorry about that: fixed

 

bobby

[Maarten Kroonenburg]

I shot a feature film this summer with the EX1. In preparation I wanted to try the external recorders, the one that was available at the rental house was the Convergent one. I shot some comparison tests and had them printed to 35mm. Then I screened them split screen with native codec on the bottom, and external recorder on top. There was absolutely no difference. Then I took the footage to the post production facility to look at was was happening close-up. As we zoomed in on the images to check color binding etc. we drew the same conclusion. So yes, on paper it sounds like it should look better, 4.2.0 against 4.2.2, 25 megabits per second against 125. But try this test and you will see for yourself.

[Bobby Shore]

I shot a feature film this summer with the EX1. In preparation I wanted to try the external recorders, the one that was available at the rental house was the Convergent one. I shot some comparison tests and had them printed to 35mm. Then I screened them split screen with native codec on the bottom, and external recorder on top. There was absolutely no difference. Then I took the footage to the post production facility to look at was was happening close-up. As we zoomed in on the images to check color binding etc. we drew the same conclusion. So yes, on paper it sounds like it should look better, 4.2.0 against 4.2.2, 25 megabits per second against 125. But try this test and you will see for yourself.

 

hey Maarten,

 

sorry for the delayed response... I wasn;'t actually able to test the nanoflash vs SxS cards before starting production. We just finished the first week of shooting, and we had some pretty major issues with the nanoflash: there were some corrupted files that had the video sync signal actually running down the right side of the frame, making the footage unusable. Luckily the footage was backed up on SxS cards, so we have the shots, although at 35mbs @ 4:2:0 instead of 140mbs @ 4:2:2 color space. Because it's so new, we weren't able to actually get an answer from the nanoflash techs as to why the footage was **(obscenity removed)**ed up... through some more research though, and talking to some other folk, it seems that it was a synch issue between the EX 3 and the NF, could've been a faulty bnc cable.

 

Also, it's been a headache for post in terms of how the clips are named by the nanoflash, and there've been a couple other issues (ie. the NF not running sometimes, and when you check playback from the camera, the NF will record it although @ 30fps interlaced, which gave us a pretty good scare when we got a call from post saying some of the footage wasn't @ 24fps progressive - luckily we were able to trace the problem to dupe clips made during playback, all you have to do is unhook the bnc cable when you playback off the camera)...

 

So yeah, definitely posing some challenges... we're thinking of ripping them off the cameras and just going on the SxS cards, but the diretor's nervous about a discrepancy in image quality on the print, so we pulled some selects (both NF and SxS) and we're screening a 35 mm print in the next couple days to see if the two will intercut... if they do, I have a feeling the NF is going to bed. Although I'm hoping there's a noticeable difference, as I'm the one who's been catching all the poop for any issues NF related.

 

What kind of tests did you run?? From all the research I've done, since the NF is still 8 bit, the most noticeable improvement in image quality will be in the movement (ie. no "tearing" during fast pans, zooms, etc)... The show I'm on right now is beyond hectic in terms of the kind of camera work, all hand held, loads of run and gun type poop, etc., the kind of aesthetic where the additional motion info would make a difference... anyways, we'll see how much production and the director will hate on me after seeing the test. I'll post the results later this week.

 

bobby

[Bobby Shore]

so, was able to take a look at the 35 film-out from both the SxS cards and the footy from the NF... vindication! after getting nothing but sidelong glances from the director and nervous conversation from the producers, the result was really pleasing: there was a noticeable difference in the footage from the nanoflash compared to the SxS. This was most prevalent in the additional motion information provided by the NF (140 mbs vs 35 mbs). In terms of the additional color depth, couldn't notice a difference (most likely cause the NF is 8 bit, so even with the 422 space, there's not much of a change from the original footage from the SxS cards 420 space).

 

That being said, footage from both could be intercut... the main difference was less strobing/tearing with the NF footage, overall it felt a bit more natural and filmlike in the motion, which was very important for this project as it has a very rough handheld feel, lots of whip pans, zooms, etc. Although this was still pretty subtle, the cumulative effect of watching a print for 90 minutes on a big screen made the choice well worth it. If I was shooting for TV release though, I'd definitely go with the SxS cards, the NF isn;'t worth the headaches we;ve been going through both on set and in post, at least for the this show. Things are going much smoother now, post has their stuff figured out and we haven't had an issue with corrupted files since swapping out the NF that seemed to give us grief during week one.

 

Three more weeks to go. Hopefully smooth sailing.

 

bobby

[Paul Kell]

hey guys,

stumbled across your post while prepping a 10 camera shoot in johannesburg. we've got 7 EX3s, 1EX1R and 2 F3s, and i'm in the DP in charge of overseeing all of the teams -- i'm looking for a picture profile (or suggestions) to get the best gradable image possible. i used to own an EX1R and had my settings as such:

 

detail -17

freq +45

crispening -45

white limit +80

black limit +80

knee apt +10

 

matrix

r-g +10

r-b +4

gr 0

g-b +14

b-r +3

b-g -3

std matrix

 

cinegamma 1

matrix +25

blk gamma -36

 

also, with the EX3 and EX1R i've shot some tests with the ninja that proved (to me, at least) there's little use for an external recorder when using these cameras since the higher bitrate does not translate to a better image. i see it as a waste of money, time and convenience. if you want the higher bitrate, just convert it to prores 422 in post, no? here's the link to my test if you want to see for yourself: http://www.eattheblinds.com/2011/05/bitrate-shootout-atomos-ninja-vs-sonys.html

 

thanks for whatever advice you throw my way and congrats bobby on Fubar 2 -- i didn't think it was possible to like it more than the first one, but i actually did.

cheers,

p