Category Archives: ExternalRecording

Quality of 4K recording

In this lesson we’ll look at two things:

  • General thoughts on the quality of 4K video from the Atomos Shogun
  • Which Prores/DNxHR version to use for 4K?

General thoughts on the quality of 4K video from the Atomos Shogun

Here are the three versions of Prores (HQ, 422 and LT) from the Atomos Shogun (click to enlarge):

Please note, these tests are mostly irrelevant because I’m using Color Finesse, which behaves strangely with Prores footage from the Shogun. Don’t use them. Skip straight to the conclusions below.

HQ2

 

422  LT

Here are my general thoughts on UHD Prores HQ:

  • You don’t have a choice if you’re recording 4K!
  • S-Log2 in Prores/DNxHR works similarly to what you get in XAVC S.
  • The resolution is definitely better, and you will get a better 1080p result if you start in 4K.
  • There are only two things important when shooting 4K: Can you handle the data rates, and is your computer good enough to handle the workload (four times 1080p)?

Which Prores/DNxHR version to use for 4K?

If you study the above images, you’ll see:

  • Visually, there is no difference between the three.
  • However, Prores HQ really shows more ‘stuff’ (data). This means, you might want to use Prores HQ for any intensive grading, chroma keying, compositing or rendering work. For DNxHR, use HQX, because it’s 10-bit.
  • For web-based work, Prores LT/DNxHD LB is more than adequate.
  • For work that needs grading but does not need Prores HQ, you will do well with Prores 422 or LT, or DNxHD SQ or LB. You can ask yourself the same questions as in the last lesson.

How to choose between Prores LT and 422 (or DNxHR SQ vs LB) for grading?

It’s about data. Here’s something to read for starters.

In short, it is not by accident that broadcast standards demand a minimum of 50 Mbps for interframe compression, and 100 Mbps for intraframe compression, for 1080p. UHD has four times the data, which means, just to match the quality of a 100 Mbps intraframe 1080p stream (Prores is intraframe), you need 400 Mbps.

This is provided by Prores LT, just as it provides 100 Mbps for 1080p. The difference in data rates between 422 and LT are not sufficiently large, but you save 30% in disk space, and your hard drives and CPUs are stressed less.

The equivalent DNxHR codec that surpasses 400 Mbps is SQ.

Therefore, when in doubt, shoot Prores LT or DNxHR SQ. If you feel a particular scene might be difficult to grade, then shoot Prores HQ or DNxHR HQX.

Click on the link below to the next lesson or head over to the main menu (above). If you need help with something, feel free to send me an email. I’ll try to get back to you as soon as possible.

My Recommendations: Should you shoot 1080p Prores/DNxHD over XAVC S?

Here are my recommendations on the 1080p Prores vs XAVC S problem. Remember, this only applies to 1080p and not UHD or 4K!

Follow this simple flowchart:

1. Are you shooting chroma keys? Or is there compositing work involved?

If yes, shoot Prores HQ/DNxHD 220x.

If no, go to step 2.

2. When grading, will I be using more than one power window or mask per shot? Or Will I need multiple renders of a shot?

If yes, shoot Prores HQ/DNxHD 220x.

If no, go to step 3.

3. Are you shooting for strict broadcast delivery (10-bit 4:2:2)?

If yes, go to step 5.

If no, go to step 4.

4. Will my work be sent to a post house or studio who will convert to Prores anyway?

If yes, go to step 5.

If no, shoot XAVC S.

5. Will there be any grading at all?

If yes, shoot Prores 422/DNxHD 145.

If no, shoot Prores LT/DNxHD 145.

6. Are you a Windows user?

If yes, stick to XAVC S.

If no, and you are a Mac user, stick to XAVC S.

Here are some FAQs:

Should you shoot Prores or DNxHD at all in 1080p?

Yes, under certain conditions.

When should you shoot Prores or DNxHD?

You should shoot Prores HQ in 1080p for the following jobs:

  • Chroma keying work
  • Heavy grading work with multiple recompressions and rendering jobs
  • Strict broadcast delivery (4:2:2 footage, sometimes 10-bit)

Should you use Prores LT or 422, or DNxHD 145?

For 1080p work, I recommend XAVC S over either Prores 422 or LT, or DNxHD 145. For the times you need Prores, you need HQ and the data cushion it provides (For DNxHD, it’s 220x).

Therefore, forget 422 and LT for 1080p work, unless you’re in a special case (see below).

When should you NOT shoot Prores?

If you’re a Windows user, avoid Prores.

I’m grading my movie or project in Resolve or Speedgrade. Should I choose XAVC S or Prores HQ or DNxHD 220x?

Choose XAVC S for all normal to medium grading work, which is 99% of projects, even cinema-quality ones.

How do you know when to use Prores HQ or DNxHD 220x? Here are some questions to ask yourself:

  • Will I use more than one power window or mask?
  • Is there any compositing work involved?
  • Will my work be sent to a post house or studio who will convert to Prores anyway?
  • Is my work intended to be delivered to broadcast (10-bit, 4:2:2)?

If the answer to any of the above is yes, then shoot Prores/DNxHD. If yes only to the third and/or fourth one, you can shoot Prores 422 or LT, or DNxHD 145.

Click on the link below to the next lesson or head over to the main menu (above). If you need help with something, feel free to send me an email. I’ll try to get back to you as soon as possible.

XAVC S vs Prores: Latency, Grading and Chroma Key

In this lesson we’ll compare the quality of the internal XAVC S codec against Prores HQ, Prores 422 and Prores LT to understand when to choose which. I’ll only be focusing on 1080p.

In this part we’ll study latency, grading and chroma key. I’m going to be comparing XAVC S with Prores HQ, 422 and LT, so we know where each one stands. We’ll be testing all of the above codecs at 24p.

Latency

There is a one-frame delay between the internal XAVC S recording and the Prores versions. This latency is obviously from the camera via HDMI – first of all due to the inherent latency in HDMI, and secondly because it has to trigger the Shogun to record as well.

In a 24p timeline, one frame is the equivalent of 42 milliseconds. In practical use, it is a non-issue.

Grading stress test

Here are full frames (JPEGS, click to enlarge) of a double-Color Finesse layer on the same Cine1 image from the previous lesson. The first is Prores HQ from the Shogun:

GradedShogun

This is the XAVC S frame:GradedXAVCS

There is only a slight difference, especially in the green channel. It is not significant. This means, for light to medium grading with just one recompression step, XAVC S is perfectly fine.

For heavier grades or multiple compression iterations, Prores HQ should be used.

Chroma key stress test (including motion)

This comparison is quite telling (click to enlarge):

ComparisonHQXAVCS

The green screen stress test is the easiest way to judge footage. You prop up a poorly lit green screen, and wave a hairy object in front of it. Then, use a really good keyer (I used Keylight in After Effects, color picked at the exact same pixel) to study the images in various stages of motion.

Here’s what the original scene looked like:

SceneChroma

The Prores HQ version is miles better in the keying department. I have included versions for 422 and LT as well, but they are really unnecessary. I have spent countless hours keying images in heavily compressed 4:2:0, and trust me, if you want to shoot good chroma key, shoot Prores HQ.

My recommendations are in the next lesson.

Click on the link below to the next lesson or head over to the main menu (above). If you need help with something, feel free to send me an email. I’ll try to get back to you as soon as possible.

XAVC S vs Prores: Resolution, Dynamic Range and Noise

In this lesson we’ll compare the quality of the internal XAVC S codec against Prores HQ, Prores 422 and Prores LT to understand when to choose which. I’ll only be focusing on 1080p.

What do I mean by ‘quality’? Here are three things we’re going to test:

  • Resolution, noise and dynamic range
  • Grading stress test
  • Chroma key stress test with motion

In this part we’ll study resolution, dynamic range and noise. I’m going to be comparing XAVC S with Prores HQ, 422 and LT, so we know where each one stands. We’ll be testing all of the above codecs at 24p in a custom white balanced setup (3100K) for the following picture profiles:

  • Rec. 709 in 709 Matrix
  • Cine1 in Cinema
  • S-Log2 in S-Gamut

Some notes:

  • DSC Labs OneShot Middle Grey was used for base exposure (0.0). It was at 1/50s, ISO 5000 at f/8
  • +3 Exposure was f/2.8, everything else as-is
  • -3 exposure was ISO 640 at f/8 for Rec. 709 and Cine1, and for S-Log2 it was ISO 4000, f/11 with an ND 0.6 (2 stops) filter.
  • In the scene, the area in the drawer (darkest) is 6 stops under
  • In the scene, the bulb is 5 stops over.
  • The scene easily has a dynamic range of 12 stops or more.
  • Even though I have tested 422 and LT, I am not including them here because it makes no difference visually.

Rec. 709 in 709 Matrix

Here’s how the Shogun compares with XAVC S in Rec. 709 (click to enlarge):

Rec709Comparison

Bottom line, there is no difference. You can use the Shogun exactly as you would use the A7s for exposure, as far as Rec. 709 is concerned.

Cine1 in Cinema

Here’s how the Shogun compares with XAVC S in Cine 1 (all Cine modes apply) (click to enlarge):

Cine1Comparison

Bottom line, there is no difference. You can use the Shogun exactly as you would use the A7s for exposure, as far as Cine modes are concerned.

S-Log2 in S-Gamut

The Shogun behaves normally. Earlier, I was seeing problems because I used Color Finesse, which is the problem, really. To know more, please read my detailed explanation in the Atomos Shogun Review.

I have deleted my earlier tests since they are irrelevant.

Bottom line, you can use the Wolfcrow system with the Shogun as well.

Results and Takeaways

  • Resolution and noise are similar between internal XAVC S and Prores
  • In all picture profiles, dynamic range is similar for XAVC S and Prores

Of course, this does not mean Prores doesn’t have its uses, which we’ll see in the next lesson.

How to expose correctly using the Atomos Shogun – S-Log2 4K

In this lesson we’ll learn how to expose S-Log2 in UHD using the Atomos Shogun.

The test is similar to what I used to explain the Picture profile modes, and I’ll be using a DSC Labs OneShot as reference. The lighting is pure halogen, custom white balanced at 2900K. The light is from the left that is why there is a small but negligible fall-off (the real world barely has zero fall-off).

The spot meter in the camera was measured within the large grey patch for exposure. Exposure settings are available in the first image.

Here are the results (click to enlarge):

SlogWCColorChart4K

Notes:

  • Look at the waveforms and vectorscope closely. You’ll see that the image in UHD has more ‘busy-ness’ than those found in 1080p S-Log2 (previous lesson – study the two side by side). Does this mean the uncompressed HDMI version has more information? We’ll look at this later in another lesson.
  • Everything else is similar to what we’ve seen earlier.

How to expose for S-Log2 UHD using the Atomos Shogun

It’s exactly the same as S-Log2 in 1080p:

  1. Use the wolfcrow system.
  2. Keep middle-grey at about 70 IRE and use 105 IRE to check for blown highlights. Use the waveform and/or zebra for this.
  3. Make sure the RGB parade lines up – this is a quick indicator of white balance issues. Forget the vectorscope.
  4. If using the false color, middle grey should be at ‘lightest grey’, and really dark skin should not fall below green. Avoid using the false color tool as a reference.

Click on the link below to the next lesson or head over to the main menu (above). If you need help with something, feel free to send me an email. I’ll try to get back to you as soon as possible.

How to expose correctly using the Atomos Shogun – S-Log2

In this lesson we’ll learn how to expose S-Log2 using the Atomos Shogun.

The test is similar to what I used to explain the Picture profile modes, and I’ll be using a DSC Labs OneShot as reference. The lighting is pure halogen, custom white balanced at 2900K. The light is from the left that is why there is a small but negligible fall-off (the real world barely has zero fall-off).

The spot meter in the camera was measured within the large grey patch for exposure. Exposure settings are available in the first image.

Middle Exposure Test

First, we’ll expose S-Log2 for 0.0, or ‘middle-exposure’. Here are the results (click to enlarge):

Slog0

Notes:

  • Middle grey falls at about 35 IRE, which is fine. The theoretical middle grey is 32 IRE for S-Log2.
  • Color information is compressed so you can hardly see it on the Vectorscope! But the RGB parade tells us the information is there.
  • The zebras are useless, because they don’t go below 50 IRE.
  • The false color tools tells us middle grey is at ‘dark grey’ (24-43 IRE) and white is at ‘light grey’ to pink (54-60-ish IRE). There’s lots more room at the top!

Wolfcrow System Tests

Now we overexpose middle grey by three stops (+3), and these are the results of the wolfcrow system:

SlogWC

The same thing but the chart is not turned for the color patches (exposure is same, nothing has changed):SlogWCColorChart

Notes:

  • The spot meter is actually blinking at +2 and is three stops over.
  • The waveform tells us middle grey falls on 70 IRE, which is where it should be for the wolfcrow system. You can see that highlights have been clipped at 105 IRE. What is instructive is that the white patch is only showing about 95 IRE.
  • Under the color chart, the black is almost touching 20 IRE, which is where it should be.
  • The darkest skin patch lights up at Zebra 50 IRE, and the darkest skin will fall at above 40 IRE, as per the wolfcrow system. This actually will help with dark skin, which was impossible to pull of with the zebra on the A7s.
  • The Zebra at 70 shows up on the grey patch.
  • The false color tool shows the lightest grey on the grey patch (58-77 IRE). However, the blacks show up as black, instead of deep blue or purple. This must be a quirk!
  • Caucasian skin can creep into the yellow zone, though I would only advise this if the highlights are not blown out.

The false color tool is not always a precise tool for exposure, as we have seen earlier. It has its uses, but is more confusing than necessary.

The biggest problem is the fact that you can’t configure the false colors in the way you want, so it becomes very restrictive. Secondly, you can’t judge blown highlights (everything’s red!) and crushed blacks (why is it showing up as black?) very accurately as well.

Underexposing and blacks

Here’s how a scene would look like underexposed:UnderExposedRealworld

And here’s a scene totally underexposed:UnderExposed

You can see that the blacks are never truly black in S-Log2, and this was explained in the guide to exposing S-Log2.

What is strange is that the false color shows up as dark blue (2-8 IRE) here, but was showing up as black in the earlier tests. It never goes purple in S-Log2, and you never have to fear crushing the blacks.

Overexposing

When you overexpose, this is what the false color tool looks like:

OverExposedAs you can see, anything above 100 IRE is red, even though the A7s can go up to 105 IRE. You can’t judge blown highlights using this tool.

However, that doesn’t mean things are bad. Here’s the actual image (full UHD 1.5 MB) of the earlier test (click to enlarge):

ShogunActualUHDImage

 

Note, some level of color compression occurs because I had to compress to JPEG to reduce space. The actual image is way ‘smoother’ and totally filmic in its transitions.

You can actually allow this camera to blow out, and it handles it exceptionally well. Compare it to the sea of red in the earlier false color tool. Totally misleading.

Are there any image quality differences between the HDMI feed and the internal XAVC S?

No.

Some users and reviewers have stated that the Prores version is more contrasty. The A7s outputs a similar signal for both, and the Atomos Shogun even displays a matching waveform.

To know the details please refer to the Atomos Shogun review.

How to expose for S-Log2 using the Atomos Shogun

Here’s the easy way:

  1. Use the wolfcrow system.
  2. Keep middle-grey at about 70 IRE and use 105 IRE to check for blown highlights. Use the waveform and/or zebra for this.
  3. Make sure the RGB parade lines up – this is a quick indicator of white balance issues. Forget the vectorscope.
  4. If using the false color, middle grey should be at ‘lightest grey’, and really dark skin should not fall below green. Avoid using the false color tool as a reference.

Click on the link below to the next lesson or head over to the main menu (above). If you need help with something, feel free to send me an email. I’ll try to get back to you as soon as possible.

How to expose correctly using the Atomos Shogun – Cine modes

In this lesson we’ll learn how to expose Cine modes in the Cinema color space using the Atomos Shogun. I’ll only be using Cine1 as reference, but this applies equally to all modes.

The test is similar to what I used to explain the Picture profile modes, and I’ll be using a DSC Labs OneShot as reference. The lighting is pure halogen, custom white balanced at 2900K. The light is from the left that is why there is a small but negligible fall-off (the real world barely has zero fall-off).

The spot meter in the camera was measured within the large grey patch for exposure. Exposure settings are available in the first image.

Here are the results (click to enlarge):

Cine1

What I see:

  • The waveform tells us middle grey falls on 50 IRE, which is perfectly fine for Cine modes. Remember, Cine modes are in full swing, so middle grey will fall on 50 IRE.
  • The Zebra at 50 shows up perfectly on the grey patch.
  • The false color tool should be showing ‘medium grey’ on the grey patch, and it does.

How to expose for Cine modes using the Atomos Shogun

Here’s the easy way:

  1. On the waveform, keep middle-grey at 50 IRE, and keep your white below 100 IRE for full swing.
  2. Make sure the RGB parade lines up – this is a quick indicator of white balance issues. You can use the vectorscope as well, but you need a clean patch to study and that’s not always practical.
  3. If using the false color, middle grey should be ‘medium grey’, and Caucasian skin could be pink or even light grey.
  4. The Zebra should be at 50 IRE for middle grey. For blown highlights, keep the Zebra at 100 IRE.

Click on the link below to the next lesson or head over to the main menu (above). If you need help with something, feel free to send me an email. I’ll try to get back to you as soon as possible.

How to expose correctly using the Atomos Shogun – Rec. 709

In this lesson we’ll learn how to expose Rec. 709 in Rec. 709 color space using the Atomos Shogun.

The test is similar to what I used to explain the Picture profile modes, and I’ll be using a DSC Labs OneShot as reference. The lighting is pure halogen, custom white balanced at 2900K. The light is from the left that is why there is a small but negligible fall-off (the real world barely has zero fall-off).

The spot meter in the camera was measured within the large grey patch for exposure. Exposure settings are available in the first image.

Here are the results (click to enlarge):

Rec709

The last three images underneath the yellow line are of the same scene underexposed by 1/3rd of a stop. I’ll explain why.

  • The waveform tells us middle grey falls on 60 IRE, where it should be lower.
  • The Zebra at 50 should show up on the grey patch, but it doesn’t.
  • The false color tool should be showing green on the grey patch, but it doesn’t.
  • This means the image is slightly overexposed.

By underexposing the image by 1/3rd of a stop, I was able to bring everything into the expected zones. The false color is green, the zebra shows up and the waveform tells us the middle grey patch is at about 45-50 IRE. White is at about 85 IRE, which is fine.

This is not a significant error, as a 1/3rd of a stop is perfectly acceptable either way. It could be the mistake of the spot meter, or of the aperture in the lens, etc. Bottom line is, it works fine.

How to expose for Rec. 709 using the Atomos Shogun

Here’s the easy way:

  1. Keep middle-grey at about 45-50 IRE, and keep your white always below 90 IRE for studio swing (this is the broadcast standard)
  2. Make sure the RGB parade lines up – this is a quick indicator of white balance issues. You can use the vectorscope as well, but you need a clean patch to study and that’s not always practical.
  3. If using the false color, middle grey should be at green, and Caucasian skin should be pink. In fact, the sweet spot is between the grays (24 IRE to 77 IRE).
  4. The Zebra is not very useful because it starts at 50 IRE and this includes the region from 47-54 IRE. It misses the green zone, so don’t use it for middle grey. Use it only to check for blown highlights, where it should be at 90 IRE.

Click on the link below to the next lesson or head over to the main menu (above). If you need help with something, feel free to send me an email. I’ll try to get back to you as soon as possible.

An Overview of the Tools Available for Exposure in the Atomos Shogun

In this lesson we’ll quickly go through the tools availablefor exposure in the Atomos Shogun .

Waveform Waveform

The waveform monitor displays IRE levels from 0 to 109, and it is the best tool to judge exposure for your camera. Even if you didn’t have access to anything else, this tool will tell you most of what you need to know about exposure.

Here’s the cheat sheet:

  • For Rec. 709 in Rec. 709 color space, middle grey should be at about 45 IRE (42-48 IRE is okay)
  • For all Cine modes in the Cine color space, middle grey should be at about 50 IRE.
  • For S-Log2, middle grey is at 32 IRE.
  • Finally, for S-Log2 under the wolfcrow system, your middle grey should be raised to about 70 IRE.

RGB Parade

The RGB parade tells you at a quick glance if there’s a color cast in your image. If all channels align up nicely, you’ve probably done your white balance correctly.

Vectorscope and Zoomed-in Vectorscope

The vectorscope is an even more powerful tool than the RGB parade. It tells you if your skin tones are spot on, and whether your white balance is spot on as well.

The zoomed-in vectorscope helps you check your white point even more precisely. This is an indispensable tool that videographers have been using for ages.

Zebra

The zebra on the Shogun is better than the one on the A7s because it extends the range from 50 to 105 IRE (on the A7s it’s from 70-105 IRE).

This will help you nail exposure on modes other than S-Log2 as well, as we shall see. The Zebra is also an excellent source to check for blown highlights or crushed shadows.

False Color

The false color tool is basically “all your zebras, but in color patches”. Here’s how the false color system is designed in the Atomos Shogun:False_Color_IRE_Small

And here’s how the colors relate to zebras:

False_Color_Percentage_Small

For Rec. 709 in the Rec. 709 color space, correct exposure will be green, and Caucasian skin is supposed to be placed on pink. In this color space and standard, you’re unlikely to use the first two and last one patch.

The A7s will not be using the last two patches in S-Log2 mode. According to the wolfcrow system, if you overexpose your middle by +3, you should see the lightest grey patch (70 IRE).

One note about the red patch: Anything above 100 IRE is in red, so don’t get scared if you see any red patches. The A7s is capable of holding detail all the way to 105 IRE. This means, the red patch is not really a good indicator of blown highlights. I’d use the waveform monitor for this instead.

Blue-only

This is mostly used to check noise. E.g., if you’re shooting at ISO 100,000 and above, this is what the normal image might look like:BluePatchOriginal

In blue-only mode, you’ll see the noise more clearly:

BluePatchNoise

That’s all for the exposure tools available on the Atomos Shogun. My picks in order of importance:

  • Waveform (most precise)
  • Zebra
  • Vectorscope
  • False Color

Click on the link below to the next lesson or head over to the main menu (above). If you need help with something, feel free to send me an email. I’ll try to get back to you as soon as possible.

What hard drives to use with the Shogun, and a data strategy for 4K editing

One of the coolest things about the Atomos Shogun, and its major advantage over the Odyssey 7Q+, is its ability to take most third-party SSDs and even 7200rpm drives.

For a full and updated list of compatible drives, click here.

The cheapest and most cost-effective drives

The choice between SSDs and HDDs are simple. Use SSDs always whenever possible! HDDs have the following disadvantages:

  • They revolve so create additional noise.
  • They get warmer.
  • They cannot be knocked about, and must remain stationary or risks failure.
  • They are slower (they max out at about 125 MB/s) and risk dropping frames.

However, I do recommend you keep one or two HDDs handy, for these important reasons:

  • SSDs are expensive, and you might run out of storage at some point. HDDs are cheaper.
  • You can record longer events with it
  • They are not bad!

I use and recommend the following drives if you’re on a budget:

  • Intel 530 240 GB SSDs (about $127)
  • HGST (Hitachi) 1 TB 7200rpm 2.5″ HDD (about $77)

Two SSDs and one HDD will run you a total of just $330 or so – that’s less than the price of one 256GB drive from Convergent Design!

If you have the money, nothing beats the Sandisk Extreme Pro series. They have proven reliable over the long term over tough professional conditions.

Data rates

Here are the data rates recording in Prores:

 x  x Prores
 x Versions 444 HQ 422 LT
1080p at 30p maximum Data rate Mbps 330 220 147 102
Data rate MB/s 41.25 27.5 18.375 12.75
Minutes per GB 0.4 0.6 0.9 1.3
Minutes per 240 GB 99 149 223 321
Minutes per 1 TB 424 636 951 1371
UHD at 30p maximum Data rate Mbps 1326 884 589 410
Data rate MB/s 166 111* 74 51
Minutes per GB 0.1 0.2 0.2 0.3
Minutes per 240 GB 25 37 56 80
Minutes per 1 TB 105 158 237 341
DNxHD
Versions n/a 220(x)* 145 36
1080p at 30p maximum Data rate Mbps 220 145 36
Data rate MB/s 27.5 18.1 4.5
Minutes per GB 0.6 0.9 3.8
Minutes per 240 GB 149 226 910
Minutes per 1 TB 636 964 3884
DNxHR
Versions HQX^ HQ SQ LB
UHD at 30p maximum Data rate Mbps 874 874 578 180
Data rate MB/s 109.25 109.5 72.25 22.5
Minutes per GB 0.15 0.15 0.24 0.75
Minutes per 240 GB 36 36 56 180
Minutes per 1 TB 153 153 245 768
  1. *220x is 10-bit, while 220 is 8-bit, that’s the difference. The equivalent of Prores HQ is 220x.
  2. ^HQX is 10-bit. It is the equivalent of Prores HQ.

For more information on DNxHD/HR, check out this white paper by Avid: http://resources.avid.com/SupportFiles/attach/HighRes_WorkflowsGuide.pdf

For more information on Prores, check out this white paper by Apple: https://www.apple.com/final-cut-pro/docs/Apple_ProRes_White_Paper.pdf

Most of the time, Prores LT, DNxHD 145 or DNxHR LB is good enough. If you’re going to do some heavy grading, then Prores HQ or DNxHD 220x/HR HQX is perfect. Visually, you won’t see much of a difference between the two, so when in doubt, use LT. You must have a strong reason to go for HQ. Read more about this in the lesson on 4K footage quality.

*Prores HQ at UHDp30 is very close to the write speed of the HGST 1 TB 7200rpm drive, so be careful!

Assuming you are shooting 4K in Prores 422, you can record up to 1 hour on a 240 GB SSD, and up to 4 hours on a 1 TB drive. If you opt for the three-drive strategy I’m following, you can shoot a total of 6 hours with the drives you have.

The best solution is to use the SSDs, and only use the HDD if you have to.

4K Editing

Every project needs 4 copies of footage:

  1. The work copy – the one you edit off
  2. The backup copy – the one you must have on standby if your work copy fails
  3. The off-site copy – the one you must have in-case your onsite copies fail, get corrupted, are stolen or get destroyed. This storage gets larger and larger the more projects you do. This is your archival system.
  4. The permanent onsite backup once your project is done. This storage gets larger and larger the more projects you do. This is your primary backup system.

For most professional-level work:

Most people combine options 1 and 2 to form a RAID array. If you’re going this route, I recommend RAID 10, with a minimum of four drives 4 TB each. RAID 5 is okay too, but I prefer RAID 10 or RAID 0.

This gives your drive system a total of 16 TB, of which only 8 TB are available. 8 TB gives you 237 minutes x 8 = 1,896 minutes or about 30 hours of footage. That should be enough for most projects. Exceptions include long-form or feature-length movies and documentaries.

A 4-drive RAID array will you have an approximate read speed of about 250 MB/s, which gives you about 4 streams of Prores 422 UHD.

Of course, you can also use a 2-drive (4 TB) RAID 0 array and copy everything to an external drive.

For 4-bay editing, you can look at these manufacturers:

Important: Don’t buy a NAS!! No Synology or WD or Netgear or Buffallo, etc.

For simple work like short films, corporate videos and music videos:

The footage needs for these kinds of projects tend to be lower, because the final output is less than 5-10 minutes on average. It is uncommon to shoot more than 4 hours of footage on such projects (though it is doable!!).

For cheap editing, I recommend the external version of the HGST drive mentioned earlier, the HGST Touro S 1 TB 7200 2.5rpm drive ($70 only).

You’ll need three of these. Two in RAID 0 and another one just as backup. Why do you need RAID 0? Even though you only need 1 TB of space, you need to edit smoothly. Most editing timelines have at least 2 tracks of video, and maybe more. The more simultaneous tracks you have (for green screen, effects, inserts, etc.) the faster your drive needs to be.

Two drives in RAID 0 will give you about 250 MB/s, which will give you about 4 tracks of UHD video in Prores 422.

If you don’t want to deal with RAID, then you can also get away with:

You can charge clients for each additional backup or copy per project. The work drives will be reused of course.

So, under this option, you can start editing projects with these drives:

  • Shogun – 2x 240 GB SSDs and 1x 1 TB 7200 rpm 2.5″ drive
  • Editing – 1x 1 TB SSD for editing and 1 x 1 TB HDD for onsite backup^ (higher if your data needs are more) and 1x 1-4 TB HDD for offsite backup
  • Total cost? About $600 all inclusive!

^You can use this drive as a production backup drive as well!

FAQs

These are important topics raised by subscribers that shed more light on this lesson.

Q. Should you worry about USB 3.0 or Thunderbolt?

A. No. USB 3.0 is fine. If you have Thunderbolt, no problem either. It doesn’t matter really.

Q. If I’m using external drives, how do I configure RAID?

A. Use software RAID. Both Windows and Macs can easily configure RAID 0 or RAID 1 without any extra gear. This is perfectly fine for most editing needs.

Q. What brand drives should I buy for RAID?

A. Hitachi, Western Digital Black or Seagate Barracuda.

Q. How do I power multiple external USB 3.0 drives?

A. Transcend TS-HUB3K USB 3.0 4-port Hub. It comes with an external DC port to provide proper power to all hubs. It can also charge an iPad.

Q. Should I use Prores or DNxHD?

A. When you have the option of both, stick to Prores. If you’re a PC user, DNxHD makes more sense. Windows and Prores doesn’t play together sometimes. If you’re a Mac user, Prores definitely.

Click on the link below to the next lesson or head over to the main menu (above). If you need help with something, feel free to send me an email. I’ll try to get back to you as soon as possible.