Mark Romero 2

47 minutes ago, KnightsFan said:

Resolve's encoders suck in general. I recently discovered that their AAC encoder produces all kinds of artifacts, so i am using trusty ffmpeg to encode audio from PCM and losslessly mux with the video.

I have not tested hardware encoding quality in ffmpeg, i use it for creating proxies so i want it to be fast rather than high quality. It would be interesting to test.

Thanks so much for the response and the information.

47 minutes ago, rawshooter said:

ffmpeg is multiplatform and only uses its own, included codec library (here is a complete list of all encoders). Performance is practically the same on Windows, Linux and MacOS, although the program is chiefly developed under Linux.

Hardware acceleration indeed degrades the export quality to h.264 or h.265 if you use the (very fast) on-chip encoders of Nvidia and AMD GPUs. The CPU-only x264 and x265 codecs of ffmpeg yield much better image quality, especially if you use the "slow" or "very slow" encoding preset and the "film" tuning parameter. (They are also available in the Handbrake GUI.) You pay with much longer encoding time. Optionally/alternatively, ffmpeg can also use the Nvidia on-chip encoder (nvenc, which you can select in Handbrake as well).

For me, as a commandline person, it's easier to just type "ffmpeg -i myvideo.mxf -vcodec libx264 -b:v 18000k -tune film -preset veryslow -b:a 192k myvideo.mp4" than clicking through Handbrake's menus, but your mileage may vary... 

Thanks so much for the elaboration.

I have one client that NEEDS the final video file to be under 500MB (as in, 499MB is fine, but 500MB is too big). It can be either h.264 or h.265 codec, and off the top of my head, the wrapper doesn't matter so much, so I think either .mov or .mp4 is fine.

The problem is, I can't "predict" how big the final video file is going to be before rendering it. So using the slow encoding presets means I have to spend an hour or so to render a three minute video file, only to find out it is over 500MB and too big for the client.

So is there a way either with ffmpeg or with one of the GUI for ffmpeg to "preview" what the final size is before rendering? Or to even limit the file by size?

Ideally, I would love for the software to just figure out what the best quality settings should be automatically while still keeping it at under 500MB final file size.

Don't know if that exists in the command line ffmpeg options or in a particular GUI.

18 hours ago, rawshooter said:

In the end, all the free/Open Source conversion programs - including Handbrake and Staxrip - are just graphical user interfaces for ffmpeg, and will thus yield the same results and performance. So one can simply stick with the tool whose user interface one prefers - or use ffmpeg directly on the command line. (Which provides more advanced functions such as ProRes and DNxHR encoding...)

Thanks for the insight.

Out of curiosity, ffmpeg is multiplatform, right?

And if so, is there any benefit to running it on a linux box over running it on a windows box?

I know Linux generally has less codex / libraries than windows but maybe there is something I am overlooking???

17 hours ago, KnightsFan said:

You can really dial in the quality/encoding time/size compromises to just where you want it, and if you can take advantage of hardware acceleration then it's lightning fast.

I've heard that - at least in resolve - that using hardware acceleration might degrade quality in terms of exporting to an h.264 codec. I don't know if this is true or not. Is there any consensus that for ffmpeg whether hardware acceleration degrades quality (or results in a larger finished file to get the same quality as using software rendering)?

16 minutes ago, currensheldon said:

I like to think that Sony has trained its customers into always needing the newest and latest tech, so they are constantly upgrading and selling every 6-12 months.

While Sony had been aggressive in their marketing, I don't really think it is accurate to say "upgrading and selling every 6-12 months." If so, we would be on the a7S V by now.

Yes, there was a case when the a6500 was released about 8 months after the a6300, but that was - as far as I understand - due to the release of the a6300 release being pushed back because of flooding at the factories in Thailand. 

Also in that instance, the a6500 was not an update or a replacement, but viewed by Sony as a higher end model than the a6300. So unlike how the a7 II was and update of the a7, and the a7 III was an update of the a7 II, the a6500 was meant to sell at a higher price point than the a6300 and for both lines to sell continuously at different price tiers.

I think a lot of the confusion comes from Sony keeping their older cameras around for a long time. For instance, a6000, a6300, and a6500 are all still for sale, despite the release of the a6100, a6400 and a6600.

But yes, it was the plan of Sony to be a "disruptor" because Canon and Nikon were so entrenched as the market leaders - and because Sony A Mount sales were so bad - that they had to be innovative and productive in their camera releases.

BTW: In case ANYONE is wondering, no, I am not advocating Sony cameras (or ANY brand of camera) to anyone.  I have a hard enough time trying to figure out which camera is right for ME, let alone someone else

On 11/4/2019 at 8:35 PM, thebrothersthre3 said:

Saw the Lighthouse the other day. Not really my cup of tea but it was incredibly well shot and acted. 

yeah... looking at the trailer... certainly not my cup of tequila.

A couple of people on the Resolve forums suggest staxrip. I haven't used it myself. But they swear by it for encoding to h.264. I don't even know if it is available for Mac.

On 1/15/2020 at 4:15 PM, kye said:

I was watching a TV show last night and they had some cool spy vs spy stuff through the streets of Hong Kong, and it was shot well, had great colour science, the story was engaging, but it felt like a home movie.  I've watched several seasons of the show and other sequences didn't have the same feel.

What's the show?

Does it have English subtitles? Or do I need to brush up on my Cantonese first?

13 hours ago, kaylee said:

first 10 sec is genuinely creepy

Oh... don't worry... it gets worse.

So who else besides me has a Panasonic S1???

And not a whole lot of love for Sony round these parts, apparently...

I haven't tested myself personally,. but I think that 1080p is supposed to have significantly less rolling shutter than 4K. (But I always shoot 4K on a gimbal with my Sony cameras, so I am not an authority on this.)

I don't think that adding sheer weight is the solution, to be honest. 

Maybe a small rig cage might help, but I think it would help by adding various grips to the cage so that ergonomically it is easier to hold. 

And the other thing is to move at the waist or at the legs, instead of moving with your hands / arms, unless you have to.

But as @mercer and @IronFilm mentioned, the Sony a6500 is not a good camera if you want to minimize RS.

2 hours ago, Márcio Kabke Pinheiro said:

Is this topic about that atrocious Northrup video, with 3 cm of depth of field, and with the image blurring in different places every 3 seconds with the poor lens trying to focus on him?

I had to use a breathalyzer after saw it to certify that I was not drunk.

Actually, no one should watch that Tony Northrup video unless they ARE drunk...

4 minutes ago, IronFilm said:

Stole it. 

https://www.facebook.com/groups/moviesetmemes/permalink/2740269309376039/

 

Nice.

52 minutes ago, Video Hummus said:

Maybe thats why f/1.2 looks so unnatural to me. See Tony's video (yeah that tony) to see why!

Better yet is Camera Conspiracies reaction to that Tony Northrup video.

I usually pronounce it "broke, eh?"

1 hour ago, kye said:

I've had success matching softer lenses with sharper lenses (or matching a softer wide-open image with a sharper stopped-down image from the same lens).

I'd recommend taking a blurred copy of the image and adding it on top with a small opacity, there will be tweaks to get it perfect but that should get you into the ballpark.

Thanks, that's an interesting though. I could probably also use something like a pro mist or netting or something??? (Thinking out loud.)

30 minutes ago, Emanuel said:

Have you taken a look on that sample up there?

Not trying to answer for @rawshooter but I liked the example that you posted. I will certainly look in to them.

Guess I was just hoping there would be some secret lens made in Outer Mongolia in the 1960's that only a few people know about and it sells on ebay for $6... with free shipping

8 minutes ago, IronFilm said:

What cameras are you shooting with currently? 

Well, right now just got a Panasonic S1 with the Sigma MC-21 adapter, so basically full frame with Canon EF glass.

The Lumix 24-105 f/4 lens the camera came with is nice except it is big and slow (f/4). It has minimal focus breathing (some say no focus breathing), and it has an automatic punch in when using manual focus with that lens, but it is kind of big and heavy, and is only f/4.

Plus it might be a little too sharp / clinical for what I have in mind.

EDIT: Most older manual focus full frame lenses can be adapted to the S1 with a simple mechanical adapter. 

2 hours ago, Emanuel said:

Samyang/Rokinon fit(s) me, no idea how it is for you, though : -)

Here's some sample, you can evaluate by yourself:

Hope this helps! : -)

Thanks. It looks pretty good.

42 minutes ago, rawshooter said:

It depends on the lens mount and sensor size of the camera you use, but Meike's Cine primes (unofficial successors to the Veydra primes) best fit your criteria.

Thanks for the suggestion.

Full frame, probably EF mount. Looks like the Meike Cine primes are for M43 and mirrorless though, right?

So are there any sleeper lenses out there for under a couple of hundred bucks that have minimal focus breathing? Preferably something around an f/2 35mm or 50mm?

(Don't exactly have to be f/2, don't exactly have to be 35mm and 50mm. Just be in the ballpark.)

Doesn't have to be super sharp. 

Nice(ish) bokeh is appreciated.

The bokeh on my Minolta MD 50mm f/2 is nice wide open, but if you stop down a little bit it (f/2.8) you get distinct hexagons. And it breathes more than I want.

Thanks in advance.

On 1/11/2020 at 1:43 AM, colepat said:

Would anyone be willing share some SLOG or HLG footage for me to mess around with grading? I want to see if I like the looks I can get. 

I will try to get you some later. 

The footage I have was mostly shot with an ND filter and that might affect the colors a bit. So will probably want to shoot some new footage for you.

11 hours ago, IronFilm said:

People are often biased towards what they own

Sure, but the main part of the test was "blind" in that subjects didn't know which brand of camera took the photos when they were choosing from a group of photos. They just saw a number 1,2,3, or 4 without knowing which was taken by which brand. 

He DID  test for bias as well and found that Sony bias was pretty high. 

And besides, the "joke" of my post was that this was a totally "subjective" test which I only brought up because the objective test I mentioned (i.e., the vector scopes showing that the a7 III color accuracy) was labeled as subjective.

Oh, and in case anyone is going to say, "But those are jpgs and we are talking about video." Yes, but VLOG plus the Sony SLOG 2 to REC 709 Gamut is meant to replicate the Standard picture profile, which is the color science in Sony JPGs.

And again, don't get me wrong. I'm not a super big fan of Sony cameras (despite owning four of them). I liked the colors out of my D750 a lot better. Just I don't think it is right to say the colors aren't accurate.

2 hours ago, Dustin said:

One thing I can’t get used to in resolve is not being able to zoom into the timeline with the trackpad like you can on PP. (I’m on a 2019 16” MBP). 

Pretty sure there HAS to be a way to do this. I'm on a PC and for that it is Alt + scroll Center Mouse Button. Certainly there has to be something on a Macbook that is similar.

1 hour ago, kye said:

and I spend 0.00000% of my life silently screaming at the camera to focus on the right thing while the moment goes by and is lost.

Your post gets a heart from me for this very line alone. Just spent about an hour out in the cold screaming at my S1 to do a focus pull on christmas lights using AF.

1 hour ago, kye said:

Not a chance.  Not even after Sony say in a press release that the entire line is cancelled.  Not even after Sony stop making cameras.  Not even after the entire company goes bankrupt and closes shop.

There are people who still think Sony will be releasing an A Mount camera any day now.

And Betamax is making a comeback... just you wait.

21 hours ago, Super8 said:

Color is not subjective no matter what he says.

Well... if it is objective, then:

17 hours ago, IronFilm said:

Once Sigma rolls out their full line up of Sigma lenses for L Mount then that will drop the costs of the L system a lot. 

Yes, but... there's still no road map for something like a 50mm f/1.8. The only thing on the road map is for a 50mm f/1.4. Sigma is probably going to charge $800+ for their 50mm f/1.4 , and the panasonic version is over $2K ?

16 hours ago, IronFilm said:

Yeah for the first year or two (or three plus...) then it doesn't really matter what you're using, as nobody is going to be hiring you likely to work within a large team where the software workflow matters. Instead you're probably mostly doing small one person jobs for indies / friends / etc.  Thus you could use Resolve, or whatever the hell you like. 

Good point.

24 minutes ago, Super8 said:

The A7III color is still off compared to Canon or Nikon.  Panasonic has improved it's color but Sony not so much.  I've color corrected a lot of Sony footage and it gets old having to save footage that was shot on a Sony.

While I prefer Nikon colors personally, I don't think it is acculturate to say that Sony color is "off." 

As per Gerald Undone, the a7 III is HIGHLY accurate when it comes to colors.

https://youtu.be/CphwVNgX32s?t=720

The Panasonic S1 needs a third-party corrective LUT to bring the colors in line.