Shirozina

2 minutes ago, Stefan Antonescu said:

A couple of years ago, I was cutting 4K H.264 files from a GH4 without any problems. Buttery smooth playback and editing.

 

1080P timeline. Intel i7 3770K, 16GB RAM, GTX 580 1.5GB. 

What NLE were you using?

52 minutes ago, dbp said:

Seems like editing is way more CPU heavy and less GPU heavy than gaming. Storage speed and RAM are also a lot more important for editing.

Gaming really has become a GPU fest. Which is nice, makes it easy in a way. 

Depends on what NLE you use.Davinci Resolve mainly uses the GPU.

DON'T THROW AWAY away your original footage as transcoding is not a perfect copy unless you use one of the uncompressed formats which will create huge amounts of data.If you transcode to a Premier / Resolve friendly format you will be looking at creating a lot of data of multiple times that of your original stock footage. The benefit of working with Proxies / Optimised media is that you can delete it after you have finished the project so it's not long term storage and with your fast CPU it won't take long to generate it again if you need to go back to the project later. Also when you come to do your final render the NLE will use the original camera footage so there won't be any danger of loosing any image quality.

Once the CPU is not having to uncompress H264 you will see speed improvements in other areas but it will depend on how much the particular process is able to use all the CPU cores. In Resolve most of the 'heavy lifting' is done by the GPU so CPU speed may not gain you much. 

Even the fastest CPU's can't handle 4k H264 yet - you will still need to generate optimised media in Resolve or use proxies / transcode in Premier but at least this process will be a lot quicker with your new CPU.

Still impressed with that score and I'm assuming cinebench is fully utilising all cores? Would be interesting to see how that CPU stacks up in real world use in NLE's which may not use all cores as effectively.

4 hours ago, Jacek said:

Sense :). I don't care if anybody agrees, just telling what I think based on my knowledge.

Overclocking is not disaster, but increasing risk of failure from for example 0.1% to 1%, which is significant, but not observed often (but still 99% of overclockers will say it's all ok).

Overclocking:

- increases risk of CPU, Motherboard failure and shortens their life time.

- increases rate of memory errors, some of them resulting in OS instability (part of Windows blue screens in old days or sudden restarts).

 

Look here (in other articles from other years you have different motherboards):

http://www.hardware.fr/articles/954-2/cartes-meres.html

Failure rate of motherboards for overclockers is many times higher in comparison to regular motherboards (despite higher price, golden pins, radiators and others). It is not because these motherboards are worse than their cheaper alternatives, they are just usually working with overclocked CPU/Memory. If you look in other articles from other months/years you can find tens of such examples.

 

I don't have time for other arguments, just do a research if you are interested. (If that's an argument :)- I'm not complete layman, have some background to judge facts - MSc in Physics and MSc in Computer Science [both on Polytechnic - more hardware related]).

 

But please avoid arguments like: My 3 PCs are working great, so i'm sure it's safe. Probably 95% or something Overclocked PC-s are working quite well, but it's not what I'm talking about.

Judge facts like this only based on statistics data, because we are talking about risk, which is statistics itself.

You have to weigh up the odds and make a choice - there's risk in everything. I wouldn't recommend someone overclock's their PC unless they know exactly what they are doing and I wouldn't overclock a CPU used in a workstation to the very limit of it's stability in voltage, temperature and frequency but neither would I state that overclocking is inherently bad or risky as a blanket statement.

What was your maximum Vcore voltage and which control option did you choose for it in the IA tweaker? Intel designs it's chips to be overclocked otherwise they would simply lock them to he stated specs like the the non K or X  models.

56 minutes ago, jax_rox said:

You can still do this. You can also add gels to your lights to get them to the right temp (as I always do with green LEDs for example).

Define 'correct' WB. Personally, I prefer to dial in my temp and green/magenta bias as I do on high-end cameras. I'm not particularly worried about getting it absolutely perfect. Don't you want to 'bake in' your desired look, rather than a 'perfect' white balance?

I assume we're talking about narrative shooting here? If you're talking events and/or news gathering, that's a different thing.

'Correct' WB is where the highlights have the same RGB values - while this may be not your ultimate look it's a good place to start and currently not easy to get to without scopes or an auto WB button.

1 hour ago, Jacek said:

Don't overclock if you don't really have to. It has big impact on stability and durability.

People are often overclocking and then complaining about instable Windows. Software is instable because of instable hardware.

Overclocking is good for gaming, but for workstation not so much.

Nonsense - nothing wrong with overclocking if done properly within the manufacturer's limits for voltage and temperature and lots to gain from faster processing. 

How can ETTR work in all circumstances with LOG? if you shoot a high contrast scene and a low contrast scene and expose both with ETTR your midtones or skin tones or shadows will be recorded at different parts of the log curve which will result in different colours (esp with Sony S-LOG/S-Gamut) and different noise levels at best and banding and other artefacts at worst. The whole point of LOG is to expose it correctly and pin your tones to the right place on the LOG curve and not squash them all up as far as they will go. ETTR is for RAW where there is no baked in response curve and you are just trying to maximise DR and where you can apply the response curve in post with no penalty of losing data. No wonder people are having problems with LOG if they are using ETTR......

LOG as used on high end cameras is not the same as the LOG you get on your sub $1000 DSLR. The former uses high bit rate, high bit depth, and high colour sampling to preserve the maximum amount of colour and tonal detail which can be easily used by even the most junior colourist/grader to good effect. The later throws away huge amounts of colour and tonal info which cannot be recovered by even the most experienced colourist/grader - banding in skies, blotchy and unconvincing skin tones, crude colours etc. You don't need a 10/12/16bit monitor to show the advantages in higher data bit rates and bit depths as an 8bit monitor will show the banding and other artifacts of low bitrate and bit depth LOG just as well as a high end Ezio display and similarly if you are manipulating high bitrate, high bit depth LOG (or RAW) on an 8bit monitor you will still be able to see how easy it is to push and pull the colours and tones and it still looks smooth and free from artefacts. Most good NLE's and Photoshop have high bit depth native working spaces so even if you are using only and 8bit image all the calculations are being done in 32bit!

IMO LUT's are being used with a lot of LOG footage to cover up it's flaws as it's easier to get a 'look' than a natural looking image with such compromised data.People are still banging on about how bit depth (10 bit is going to fix everything......) is the problem with LOG on DSLR's when the reality is the high data compression used in internal codecs. S-LOG2 8bit HDMI from my Sony A7s and A7r2 to an external recorder is very flexible (within reasonable limits) whereas you can often see banding on even unedited footage from internal S-LOG. Also exposing LOG properly is not easy as you need good scopes to pin the tones to the correct exposure level. These scopes are only available on external recorders or high end cameras so your are at another disadvantage shooting LOG on your DSLR. Many just use ETTR but this is exactly the best way to get into problems on low bitrate codecs as data compressed too far up the log curve is then dragged back down in grading where it falls apart - skintones, skies etc.

Getting it right in camera is no that hard and most non log camera profiles have enough adjustability to set your contrast appropriately to the scene you are shooting and are not far off LOG in terms of DR - not shooting LOG can be very liberating! 

45 minutes ago, Santiago de la Rosa said:

Ryzen 1700 mounted, right now overclocked to 3,7Ghz all cores. 35ºC idle, 50-55ºC rendering. It´s really faster than my i7 2600K overclocked to 4,2Ghz.

1610 Points Cinebench R15. I will be editing next week with a new project, more conclusions to come.

I´m very satisfied.

Noctua NH-U12S AM4

Gskill tridentz 3000-cl14 2x8gb overclocked to 3200 cl14

Ryzen 1700 overclocked to 3700Mhz all cores

Asus X370 Prime Pro bios 0614

Impressive - I can only get a cinebench score of 1330 with my i79850k OC'd to 4.4ghz

9 hours ago, dantheman said:

If you want easy video functionality buy a professional video camera that's geared towards video only.

Thanks for the advice - any particular models you can recommend?

12 minutes ago, Kubrickian said:

Huh? Tell that to all the Super 16mm filmmakers out there. Jackie with Natalie Portman was shot on 16mm last year buddy. If you have a good story to tell, you could shoot it on any of the cameras spoken about on this forum. Stop rationalizing your own creative deficits. 

I was being sarcastic.......

Fd lenses are ideal for the Sony A7 series although you have to choose your glass carefully as some are better than others. All of the fixed lenses are good whenstopped well down and have very good frame evenness I.e sharp right out to the frame corners unlike a lot of similar vintage and even contemporary lenses. Wide open they are mostly low contrast and ' dreamy' but this can be attractive but Bokeh can be harsh and so they make good ' landscape lenses' Outright resolution is not exceptional so I would caution their use on high density aps-c and M43 sensors? Most of the zooms are mediocre by modern standards ( as are any of that era) Any L glass is good but they are command high prices and are not such bargains. They are plentiful and cheap on the used market due to their non compatibility with the EOS mount. 

My xc10 has a constant aperture when I set it to 5.6 or smaller - shallow DOF is not the answer to all situations. But - it's parafocal, has a very good range, is small and has a very effective image stabiliser all of which may have had to be compromised to get a constant 2.8 aperture. It's biggest let down for most though I suspect is that it has a 1" sensor which mean's it can never be 'cinematic'.......

I meant a custom WB rather than just a button to access the WB selection menu.

4 hours ago, jax_rox said:

Given that modern SLRs/SLMs and large sensor video/digital cinema cameras are digital attempts at mimicking the film cameras that came before - it makes absolute sense to me.

There was no 'custom white balancing' off a white card in the old film days. Such a process was reserved for a small subset of video cameras. You might shoot a grey card and/or colour chart on a roll of film, but the colour temperature was (and still is) 'baked in'.

To that end, the fact that you can dial in a custom WB temperature is significantly more advanced than what used to be. As custom WB off a card is more of a ENG/Doco/Corporate thing to do, it doesn't bother me that it's stashed away in a menu somewhere. I'd rather dial in the temp anyway.

You can't WB off a white card on an Alexa... 

Back in the days of film you used a colour temperature meter and then added colour correction filters to the lens. If you shoot RAW video  you don't need to WB as you can fix it in post just like shooting RAW stills. With an 8 bit 4.2.0 codec you need to bake in the correct WB and it's a very easy to do this with a custom WB feature. When I have my external recorder hooked up I can get very close with the scopes and adjusting the K temp and Green / magenta axis.

Shooting with a low bit rate and depth internal codec it is essential to nail the correct WB in camera so why don't any cameras have a simple function whereby you can assign a custom button to set WB by pointing it at grey card/expodisk etc? Sony A7 series being the worst as you have to change away from movie mode (still an ongoing WB bug in all their cameras in LOG) then dig deep into the menu structure to find the function which should (could with a simple firmware change) be just a custom button away. I seem to remember Canon being a bit easier and some cameras not even offering this at all. Seems like a basic control to me like being able to set your aperture, shutter and ISO etc.....

4 hours ago, BenEricson said:

Depends what you're doing. If you're shooting a 4k landscape, this actually a better camera than a lot of other options. Green actually looks like green. Skin tone is the best as well. Built in lenses have their perks. 

Loads of people are buying the FS5, but the c100ii puts out a much nicer looking image. It all depends what you're doing. 

Im not sure why people write off color as not being a feature. ?

All of this makes me think the 1DC is probably the smart buy. 

I found the log to work nicely if ETTR. You also need to sharpen in post because C-Log has no sharpening. 

Have you tried WIDE DR with the new firmware? Does that have noise reduction applied baked in? 

LOG + ETTR 'was' the problem for me - sky tone pushed too far up the curve then falling apart when pulled back down the slope in grading. Even when exposed properly for LOG ( i.e not ETTR) it would band like hell with the slightest tonal adjustments. I've just loaded up the latest firmware so I need to test all the profiles again even though Canon hasn't stated they have made any major changes to image quality. It was obvious from looking at footage with a lot of fine detail  that there was a lot of NR going on even at native ISO's.

I'm giving mine another chance by abandoning shooting in LOG. Already the colours are a lot better and the mushiness in fine detail is gone. So far I'm using the profile 4 with sharpness and contrast down a notch or 2.Will try and find some blue skies and see if the banding has improved........

I have a small collection of cans; Sony DMR 7506, Sen HD25-2, AKG 701 and Grado SR80. If I want to listen to music I'll grab the Grado's as they are sweet sounding but not very accurate - classic Hi-Fi vs sound production choice. If I was forced into mixing with headphones I'd choose the 7506's - no they may not be the sweetest sounding or most detailed but like studio monitors if the mix sounds good on these it's a good mix. For recording in the field the HD25's are unbeatable but not particularly because of the sound but because the split headband means they don't slip off your head when you look down at the field recorder and they are tough and comfortable to wear for long periods. Bought the AKG's on a whim but the best thing I can say about them is they sound inoffensive. I know there are some fantastic sounding cans out there but for mixing and sound production this isn't actualy that useful. NX10 monitors became studio essentials not because they sounded good but because only a good sound mix worked on them and it ruthlessly revealed poor production - if your mix sounded good on these it would sound good on anything. I'd think you need the same from cans and the 7506's do this as well which is why 1000's of audio pro's still use and trust them.

No it's can't be binaural as it's very directional but at the same time it has a wide soundstage.....

Colour and contrast may be a bit too strong for some but it can't be said that they are not clean and accurate. The sound is very good - binaural?

Extreme narrow DOF has very little use if you are actually shooting a film as opposed to posting clips on the internet to make people go WOW! Most of the time you need more not less DOF. It's not like the laws of optics are new and still being worked out so why are people still arguing over 'equivelence' and some magical reality bending properties of MF lenses......