Sony A7s' HDMI output destroys SLOG-2 gamma curve!

[Cliff Totten]

Hi folks,

 

As far as I know, very little has been said about this. However, Convergent Design has recently stated that Sony has the A7s HDMI port locked to a gamma curve that is closer to REC 709 than true SLOG-2.

 

It's true, I have tested this myself and the A7s destroys it's SLOG-2 gamma curve on it's HDMI port.

 

Here is what happens:

 

1.) Record a SLOG-2 internal XAVC-S file.

2.) Record via HDMI to any recorder that you have.

3.) Compare both files on a scope...you will see that the SLOG-2 gamma curve has been stretched and "damaged". (compared to internal file's "proper" curve on the scope) 

 

The histogram does not lie either, it CLEARLY shows what Sony is doing to cripple it's HDMI port. (hopefully this is a Sony oversight and not a "surprise...gotcha!" stunt.

 

Sony markets its A7s external recording HEAVILY against the GH4 and it's a big selling point for the A7s. SLOG-2 in general is a HUGE selling feature of this camera. Now, if Sony chooses to cripple SLOG-2 over the port and trap it inside it's internal 4:2:0 codec,...that is perfectly well within their right. (No doubt at all)

 

However, Sony needs to state this is being done in it's marketing and sales materials!! Sony needs to be upfront about it when they push it's external recording capability!

 

Statements like:    * "SLOG-2 has been removed from any external recording"   or  * "What is displayed on the camera monitor (gamma curve) is NOT what is actually output over HDMI.

 

This would be the FAIR way to treat Sony's customers.

 

This is a huge problem and I hope Sony fixes this soon.

 

Am I the only A7s owner that is concerned about this?

 

 

 

 

 

 

[Guest Ebrahim Saadawi]

Did Sony ever say the S-log 2 will be output via the HDMI? Curious. It should. Specifically for a camera that only records its best mode via HDMI.

[wjkotze]

Do you have a link to the Convergent Design statement ?

[Andrew Reid]

Let's take this with a pinch of salt until someone impartial has tested the Shogun with the A7S.

[Cliff Totten]

Let's take this with a pinch of salt until someone impartial has tested the Shogun with the A7S.

 

Absolutely agreed! I'm desperate for more A7s testers. The original notice went up on DVInfo. I saw it and began testing myself. I tested with both Atomos and BlackMagic recorders and both gave the same result.

 

The gamma curve that is recorded internally via XAVC-S is very different than what is sent out of HDMI.

 

Compare both files (XAVC-S and ProRes file) on your scopes and tell me what you see.

 

I hope you don't see the same thing that Convergent Design and I saw.

 

Calling all A7s testers!..please.

 

I would sooooo very much LOVE to be wrong about this.

 

CT

[Guest Ebrahim Saadawi]

Anyone who has the camera, set it to S-log2, connect the camera via HDMI to a TV, and look if the image is S-log2 or crushed compared to the one on the camera screen.

Lets leave recorders and competition out of it and just see what the camera output looks like. It would be a shame if it's crushed with an S curve instead of true S-log2. 

[Cliff Totten]

Anyone who has the camera, set it to S-log2, connect the camera via HDMI to a TV, and look if the image is S-log2 or crushed compared to the one on the camera screen.

Lets leave recorders and competition out of it and just see what the camera output looks like. It would be a shame if it's crushed with an S curve instead of true S-log2. 

 

It's important to note that when using SLOG-2 on the A7s, there IS a "LOG-ish" type of gamma curve. However, it is NOTHING like true SLOG-2.

 

The HDMI port seems to rest somewhere between REC 709 and SLOG-2. You must compare with scopes. Histograms don't lie. You will see a SIGNIFICANT amount of contrast over the HDMI signal vs. the true SLOG-2 internal recording. It's my guess that you lose between 1 and 2 stops of dynamic range over the HDMI. (This is a rough guess of mine.)

 

The bit distribution is WAY off between the two. It's very easy to see.

 

Dan Keaton from Convergent Design has stated today that they are calling their contacts at Sony about this after IBC is over. They don't feel this is intentional on Sony's part.

 

Again, if you are an A7s owner and can screen cap your scopes and histograms of the two files, (HDMI vs. internal XAVC-S) that would really go a long way to help us get Sony to fix this. (see mine above)

 

CT

 

Edit:  BTW,....this is NOT a recorder problem. We have confirmation from users stating that Convergent Design, Atomos, BlackMagic and PIX recorders ALL see the A7s HDMI this way.

[Andrew Reid]

Edit:  BTW,....this is NOT a recorder problem. We have confirmation from users stating that Convergent Design, Atomos, BlackMagic and PIX recorders ALL see the A7s HDMI this way.

 

I thought the 7Q didn't accept 4K via HDMI from the A7S at all, which is why we're all waiting for the Shogun...

 

What's changed?

[Cliff Totten]

It happens at 1080 HDMI capture. I have not heard of any reports on the 4K side yet. (Dont know how many people are actually recording 4K form the A7s today)

 

I hope Sony fixes this before the Shogun comes out. The A7s relies heavily on it's HDMI port to compete against the GH4 and I'm sure this news will become bigger and bigger after the Shogun is released. (There will literally be thousands of people recording 4K form the A7s by November)

[Tim Naylor]

Does this gamma cripple happen if you go through a 4k - 1080p convertor - to Atomos?

[Cliff Totten]

Does this gamma cripple happen if you go through a 4k - 1080p convertor - to Atomos?

No way to know yet about the 4K output. I'm going to hold back on calling it a "cripple" right now. I'm just going to call it a "Sony oversight" today.

 

If they fix it than it's just s simple design mistake. If they decide not to fix it? Than for sure, it's a Sony "cripple".  It is quite possible that they don't want the A7s being able to output SLOG-2 from HDMI to a higher codec like ProRes (especially in 4K) because they find that too dangerous to their other models. Yes, it could be a sneaky and surprise "gotcha" for their customers. (A very bad PR mistake if so)

 

Let's wait and see. If Sony fixes it than all will be forgiven quickly, I'm sure.

 

It's been reported to Atomos and their engineers are investigating. Convergent Design is checking with Sony.

 

We need more A7s testers!!!

 

CT

 

Let's see what Sony says.

 

CT

[Andrew Reid]

Very interesting, thanks for the info Cliff. Keep us updated.

[hmcindie]

Sooo... what happens if you record without s-log? Just a regular picture profile or with one of the cinegammas? 

[Viet Bach Bui]

This video from a while back has the internal vs Pix240 comparison:

 

Cine 4: seems to be the same.

S-Log 2: there is a difference, but nothing significant in my opinion. I actually think the Pix240 footages would be easier to grade.

 

[Cliff Totten]

You can see at 3:44 that the highlights in the lamp are smoother with the internal XAVC-S and will eventually avoid clipping better under "true" S-LOG2. 

 

When you look at the two captures on a scopes, there is a very significant difference between them graphically.

 

So far, I think we can say that what is being output over HDMI is NOT S-LOG2. It's a gamma curve with about a stop or more LESS dynamic range than "true" S-LOG2.

 

Looking for more A7s testers out there.....

 

CT

[Cliff Totten]

I don't know if I'm not allowed to post test results from another site but here goes:

 

http://www.dvxuser.com/V6/showthread.php?328412-Sony-A7s-DESTROYS-it-s-own-SLOG-2-gamma-curve-over-it-s-HDMI-port&p=1986476402&posted=1#post1986476402

 

Another test from another A7s owner that displays this same problem,....

 

Any more A7s testers out there? 

 

CT 

[hmcindie]

This could be a superwhite problem. Have you tried adjusting the balance afterwards, seeing if you could bring possible superwhites back? In Premiere CC you can do this with the fast color corrector (set it to 16-235 instead of 0-255) or in AE in 32bit float.

 

the avcx-s slog2 doesn't use superwhites (cinegamma4 does) but it could be that the NLE doesn't see the superwhites properly. Can you try getting them back?

 

According to Paul in the dvxuser forum, that's exactly what's happening so basically your not really losing anything (same thing happens when recording cine4 in avcx-s, you have to bring the superwhites back later (and the blacks too)).

[Cliff Totten]

Here is what people have determined so far:  The A7s internal XAVC-S seems to use the 16-235 range and avoids exceeding the upper and lower end of that scale. However, it also seems that this is not so with the HDMI output. Values for that WILL exceed 16-235. The A7s adds contrast to that to bring those values to 0-255!

 

Its possible that it's not losing any image data by doing this. It simply maps "16" to "0" and "235" to "255" on the HDMI output.

 

This will certainly cause a noticeable (and alarming) difference in brightness between XAVC-S and HDMI recorded images. However, it might be OK because no real image data is lost in that process. (theoretically and hopefully)

 

People are still trying to determine if the actual SLOG-2 gamma curve is maintained in this contrast spread.

 

Love to know more of what other A7s testers think about this.. 

 

CT

[IronFilm]

Last update to this was a month ago, wondering if any progress/news has happened recently? (especially from Sony, or firmware updates)

[sunyata]

This could be a superwhite problem. Have you tried adjusting the balance afterwards, seeing if you could bring possible superwhites back?

 

I was thinking the same thing, the histogram looks like a scaling problem. Has a test been done to make sure that just scaling the gamma into the correct range won't solve the problem? a different shaped gamma curve when using HDMI seems unlikely, but who knows... a custom 1D shaper lut might be a fix if scaling is the problem.