Gemini 444 F3 and S Log

[Ben Peake]

hi at the risk of sounding like a real dumbass i gotta ask this, i have looked high and low for information on S Log and 444 editing and i cant find anything anywhere,,, so has anyone got the answer to editing with s log in 444 or s log in any rasta if that works,,any information out there search engines come back with nothing for me
thanks in advance

[Nate Weaver]

Originally Posted by Ben Peake

hi at the risk of sounding like a real dumbass i gotta ask this, i have looked high and low for information on S Log and 444 editing and i cant find anything anywhere,,, so has anyone got the answer to editing with s log in 444 or s log in any rasta if that works,,any information out there search engines come back with nothing for me
thanks in advance

What do you want to know about it?

Gemini will record uncompressed 4:4:4 Quicktimes, which you will need fast drives for unless you convert to ProRes or the like. Otherwise, FCP will have no problem with these files.

You will likely want to apply an S-Log LUT to the files as you edit, just so you can see a better looking picture. Couple ways to do that:

1-There will likely be an FCP plugin to apply the LUT, and make things nice via a filter. One exists for the Alexa Log-C, and it showed up fast after the Alexa hit the streets.

2-various companies make black boxes to put in line on the SDI output, to apply the LUT to the video out signal before it hits your external SDI monitor. Think Blackmagic, AJA

3-Apply a 3-way color corrector filter in your NLE, apply some contrast and saturation, bring down the gamma and tell everybody it's not a final color look. Remove these filters before output for real color correction work after edit is locked.

[Doug Jensen]

A better way is to record simultaneous 35Mbps footage on the SxS cards with a LUT baked in. The F3 will have that ability.
Then use those clips as proxies for the S-LOG files. Ingest and edit with the proxies, and then conform at the end of the edit and replace the proxies with the full-res S-LOG files.

[Nate Weaver]

Originally Posted by Doug Jensen

A better way is to record simultaneous 35Mbps footage on the SxS cards with a LUT baked in. The F3 will have that ability.
Then use those clips as proxies for the S-LOG files. Ingest and edit with the proxies, and then conform at the end of the edit and replace the proxies with the full-res S-LOG files.

Just a note Doug, don't want to be argumentative, I promise...

I've done a lot of Red conforms using ProRes transcodes, and other ways as well. There's a LOT that can go wrong and cause huge stress headaches. The phrase "just conform at the end" is anywhere from 20 minutes of work to a few days, depending how detail oriented people were on set. Or in the case of Red, how careful they were with their transcodes. Slapping a FCP filter on (assuming drive throughput is in place) is a LOT easier than any offline/online conform with media coming from two different places.

If your editor goes to town without making sure the 2 sets of media match perfectly, you're in for a world of sorrow.

I'm a little concerned, as somebody who does post, about the idea of clip numbers between a Gemini and internal SxS clips names becoming unsynced with this method. I think it can of course be done. The official Sony method, using a SR-R1 recorder I would imagine will auto sync clip names and metadata between cam and recorder, making this a lot more foolproof. So I guess all I want to do here is warn people who haven't done it before it can cause a lot of time/overages if you don't watch your Ps & Qs.

I'm thinking in the next 12-18 months, uncompressed 1080p is not going to be a big deal at all, ESPECIALLY with Thunderbolt this and that coming down the pike.

[Charles Papert]

Nate, do you feel like a lot of folks are going to start waving the 4:4:4 flag as the latest flavor of "must have" who ultimately don't really need it? Having gotten used to the concept of 4:2:2 as being "good enough" for network television acquisition in recent years, with 4:4:4 reserved for theatrical features, it seems to me that for the majority of readers of this board, 4:2:2 should probably suffice. I know this is a subjective thing, but given a project that will likely not make it to theatrical projection (or even if it does), what percentage of improvement would you place on 4:4:4 vs a good 4:2:2 codec?

[Doug Jensen]

Originally Posted by Nate Weaver

Just a note Doug, don't want to be argumentative, I promise....

Nate, I'm on your side. I'm just stating how it is "supposed" to work. Until I try it myself, and see how all the pieces fit together, I won't be a believer either. I'm skeptical of everything, until I see it myself.
In the mean time, I'll defer to your judgement since you are more experienced in this area, but I do have my fingers crossed that my first post will prove to be the reality.

[Nate Weaver]

Originally Posted by Charles Papert

Nate, do you feel like a lot of folks are going to start waving the 4:4:4 flag as the latest flavor of "must have" who ultimately don't really need it?

Speaking as a colorist, not as a shooter (oh man, multi-hyphanate overkill)...

4:4:4 is overkill for anything but keying and stuff that will see lots of FX manipulation. For garden variety color-correction, 10 bit and/or uncompressed is MUCH more important. I can't stress it enough.

10 bit allows a colorist to change black level/exposure/gamma without posterization.

Uncompressed allows a colorist to pull keys for secondaries that don't get blocky (you can see the compression macroblocks show up a lot more the heavier the compression is)

[Nate Weaver]

Originally Posted by Doug Jensen

Nate, I'm on your side. I'm just stating how it is "supposed" to work.

Right. I hear ya. And anybody asking the question here should be aware there's another way I didn't mention.

If it can be done relatively easily, it will be indeed a great way to go.

Davinci/BM Resolve, the way it conforms, would actually tackle mismatching clip names like a champ, since ALL it does it look at reel numbers and timecode to assemble a conform...it doesn't care about clip names or even how many files that footage comes from. That approach has some big downsides too, though. All 5D footy has to have TC or you're skee-rooed in the conform.

Here's hoping the new FCP has some smarter/new ways to reconnect media to help out future file based flows!

[Ben Peake]

thanks ever so much everybody thats really wonderful and helped me understand a bit more,, nate that daVinci looks so awesome a program,, do you find your at the end of the line though and the moneys been spent by the time they get to post,,?
im in Australia and they have spent the budget before they get to shoot lol,,
seriously in Australia multi national mega rich companies have $2000 budgets for commercials,, its a crazy world
thanks again

[Ben Peake]

i should have mentioned that for most part the footage from the cards at 4.2.0 would suffice for the majority of my work, (extreme low budget) but its nice to have the options and understandings for any future reference, thanks again

[Dave Sperling]

Originally Posted by Ben Peake

do you find your at the end of the line though and the moneys been spent by the time they get to post,,?

Running out of money can be a problem for just about any production; however, S-Log won't help you much if you can't afford the conform and color correct!

BUT -- the way I've been interpreting the product notices, the firmware upgrade that enables S-Log not only gives you four pre-defined LUT's to choose from for the data being written to your SxS card (and output through your standard HD-SDI port), but also allows you to create several LUT's of your own. What that means is that you would be able not only to record in S-Log (to a Sony SR-R1 or C-D Gemini or similar using the dual link connectors), but also record to your SxS cards (and if you wish, to an outboard 4:2:2 recorder using the regular HD-SDI port as well) using a LUT that you have set up to get an image on set as close as possible to what you are envisioning for the final color correct. Thus, if you run out of cash and can never afford to go back and use your s-Log data, you at least have a good looking image to work with / show / and even use as your final if necessary.
A typical scenario might be to shoot a low budget feature to both a NanoFlash (at 100+ Mbps) and an S-Log recorder simultaneously, edit using the LUT-baked-in NanoFlash footage, and use your NLE to output a screenable version for festival submissions and distributor screeners. Then, when you get your 'big sale' or that Sundance acceptance, actually spend the $$$ to do the conform and correct from the S-Log for optimal quality.

[Doug Jensen]

Quote: "Thus, if you run out of cash and can never afford to go back and use your s-Log data, you at least have a good looking image to work with / show / and even use as your final if necessary."

---

Not so fast. I'm not so sure that will be possible. I haven't tried the S-LOG workflow myself, but my instincts tell me that you will NOT be able to set the camera up so that you can cover both bases at once. You're either going to have to set the camera up and expose for S-LOG . . . or for REC709 or at least one of them will be compromised. I really doubt that you're going to be able to sit on the fence and make your choice later in post. Even considering the use of LUTs and all that stuff, the two shooting modes will surely have two different exposure requirements.

Just look at the Las Vegas S-LOG demo footage that was shot during NAB to see what I mean. It is ridiculous to think that that test has any value at all. It demonstrates nothing. If the camera is set up for S-LOG, then REC709 is compromised, and vice versa. A more valid test would be to have two cameras, each set up perfectly for either S-LOG or REC709 . . and shoot the same shot with both cameras to compare the results. That was not done on that Vegas footage. I know without a shadow of a doubt that my F3, with my personal Picture Profile, would have looked better than all of the footage from that test labeled "XDCAM" and even some of the S-LOG graded footage.

So, I'll leave the door open for maybe changing my opinion later after I get a chance to play with the whole S-LOG workflow, but I'm willing to bet that you're not going to be able to wait until post to decide which one you're going to use.

[Alister Chapman]

I'm with Doug. The LUT's are there to give you an approximation of what the graded material should look like. I don't believe you can seriously shoot S-Log and then expect a baked in LUT to give you results that will be as good as what you would have got if you shot well exposed with Rec-709 or a Cinegamma/Hypergamma.

You would be better off IMHO shooting S-Log and working with S-Log. Applying a curves or S-log plugin filter to your footage will not be particularly hard or time consuming, no worse than grading Cinegamma or any other footage. The first few jobs may take a little longer while you work out which filters or plugins to use, but once your up to speed it should not be a particularly hard or time consuming process.

The difference is that with a standard gamma you don't have to grade at all, but with S-Log you must grade. However the reality of any kind of feature or other quality production is that you would almost certainly grade anyway, so S-Log really isn't going to add any significant cost or time IMHO as it is just a grading process.

It will be interesting to see how this goes as this is the first time anything like S-Log has been available at this cost level. Just how far down the food chain will S-Log creep??

[Dave Sperling]

Hello Doug, Alister,
I certainly hadn't spent much time thinking about the relative exposure levels between S-Log recording and standard recording to the cards, so you may be right about compromising the image recorded on the SxS cards when exposing properly for S-Log. Since the ISO rating of 1600 for S-Log compared to 800 for Rec-709 would tend to indicate that the SxS footage would be underexposed by one stop, I'm guessing it will be up to the LUT to raise the exposure level recorded on the SxS accordingly.
On the other hand, I'm still intrigued by the possibility of creating a number of user-defined LUT settings, and until we see an initial non-beta release of the new firmware I'd like to reserve my final judgement.

Doug, I agree that the 'demo test' footage we saw from LasVegas didn't really prove anything because it's correcting the s-log footage and not the SxS footage - plus we're not sure how the exposures were set.
However, I think your comment about Picture Profiles opens a whole other line of discussion - since I'm wondering if you'll be able to create a LUT version of your favorite F3 profile that can be used while recording s-log -- or whether the user defined LUT will not give us that much adjustment room. I certainly like what you've done with your F3 profile, and I'm looking forward to seeing what you can do with the user LUTs.

Alister, having shot several low budget features that have (unfortunately) at various times run out of funding money (only to find new investors/ raise more money / and eventually be completed and released) , I'm not suggesting a theater release version that uses the SxS footage rather than corrected S-Log footage, but I know that often the producers may need to screen whatever they have available, particularly as part of raising additional funds, or submitting to festivals, or pitching to distributors. For these purposes it will often not be affordable or viable to go back to S-Log data and color correct, so having the best looking LUT footage possible could make a big difference in getting the project completed.

Best,
Dave S.

[Leonard Levy]

I'm confused by a lot of the Slog discussion and probably it won't all become clear until I actually see it, but here's a few questions to start with:

How will the LUT's work when shooting SLog?

Will you be able to see them when monitoring? Will you need a separate black box to do that after the recording stage? (one more expense & hassle).

Does the SxS card have to record 709 (presumably at the wrong exposure) when the SDI is spitting out SLog or do you have the full set of gamma menu options? Might one of those options be the SLog with LUT applied?

Will it be possible to adjust the LUT scene to scene through the picture profiles?

Both Alister and Doug say that the SLog with LUT applied still won't be as good as a carefully recorded 709 or Hypergamma, but would it be as good as a RED with Rec709 or Red Space look applied.

It would be nice if the monitor output could give you the LUT applied ( while SDI sent out the Slog) and you could set up a switch with one of the user assign buttons to jump from Slog to LUT similar to what you have on a RED. (Likewise the same switch to the cameras own LCD & EVF.)

Red for all the headache of its workflow allows a very easy way for the client to decide if he wants to bypass the Raw recording and use a Proxy to cut fast. It would be sad if the heralded easy workflow of the F3 was worse.

I agree with Doug about not learning anything from the Las Vegas tests. They showed that SLog looked good but the comparison was pointless and only made the 709 recording look like junk which its not.

I hope Sony really spends time with this to give us a flexible recording system. Some the upcoming firmware changes show that they're listening.

One thing I would like to see is a much wider array of choices for the user assign buttons. Even say a few fixed Kelvin choices like 2900, 3200, 5600 and 6200, or adding the audio wind filter would be huge time savers.