Scott_Warren

Here's the folder of ISO 800 flickering frames of a colorchecker chart:

https://drive.google.com/open?id=1YLY5ls433I-DgaLSbbzE3Lf6MtTKwjKT

@Lars indeed! When I used 172.8 the flickering also occurred. I'm not certain that the behavior is affected by the designation of degrees or speed in my tests so far. 

21 minutes ago, paulinventome said:

That could be compression though...

What shutter speeds are you using and how do you specify them, in degrees?

cheers
Paul

I'm using 1/50th shutter specified by speed, since when I first got the camera it seemed that leaving it in shutter angle of 180 with the framerate at 23.98 still created files with 1/50th shutter speeds instead of 1/48th. Just tried the test with the colorchecker (uploading the folder now) and switched to 1/60th briefly in a separate test and the flickering still occurred. 

Another sample of flickering. It's subtle, but check out the left part of the frame here beginning at 17:00

It even happens in the next shot recorded in .MOV.

Cue the dramatic conspiracy music. 

24 minutes ago, Chris Whitten said:

How do I check for flickering in my own footage - what is the methodology?

Forgot to mention, I'm also on 25fps.

I bought and am using the Sigma LVF.

The diopter moves as you walk along, which is annoying. It takes away some of the grip points on the camera, so it is harder to hold.

On the upside, it helps me focus, compose and check exposure. It is very light, but it more than doubles the bulk of the camera.

Heya Chris,

Here's what I do. Capture at 4K in either 12bit/ 10bit. For every ISO starting at 100, I capture 10 seconds of a static shot. Seems like a long enough time where you'll see it happen if it's going to.

Then fire it up in Resolve and add each clip to a new timeline and look for the waveform behavior seen here:
https://drive.google.com/file/d/1kMpU9QFj92Ly-h7aOkN5Asb9ombpMolX/view

You might not see it in the footage, but the waveform always tells the truth

Paul, I do have a colorchecker! Are you saying to lock off a shot of the chart exposed at speed?

1 hour ago, redepicguy said:

 

That's crazy! I'd love to know how widespread this actually is. Most people will use it as normal (likely with .mov capture) and roll with it without doing pixel peeping so they might not see it, but surely it's not just a handful of cases.

1 hour ago, paulinventome said:

Ah, well yes, do that. Because i find it difficult to believe that the physical sensor or hardware is changing depending on ISO like that and the way the data is looking it seems like a bug perhaps between sensor and writing out the RAW data. 8bit and 10bit would be slightly different.

I've mostly *not* used 12 bit because i am in 25p land. I have used it when shooting without sound sync but certainly beyond tests and messing around i've not 100% focused on it...

cheers
Paul

Drumroll....

10bit has the same behavior as 12bit  Though it seems like 400 is actually now stable with 1.02. No flickering with 12bit or 10bit after checking yesterday's stuff as well.

100-400 = fine
500-2000 = flicker
2500 = fine
3200-4000 = flicker
5000-6400 = fine

Secondary observations:
The over-exposed first frame issue happens only on 125-320. Stable for all others. Possibly related to ISO scaling down from "native" sensor values somewhere? It's interesting that ISO 100 & 400+ is stable and behaves as expected.

For now, I'll just treat the camera like a 400/2500 dual gain kind of deal until I can send it to Sigma for diagnosis. 

 

23 minutes ago, paulinventome said:

Initially it reminded me of using canon lenses many years ago where the electronic aperture was not 100% accurate from frame to frame. But then further investigation shows that this just appears to be in the shadows. (Although it could just that it is more observable in the shadows). The two screens show the shadow data changing quite obviously. There are two 'modes' that it appears to be flicking between

It doesn't appear to be an overall exposure change.

So is this identical in 8 bit, 10 bit and 12 bit? Because one finger could point at the encoding. If it's the same on all of those i would have to suggest the body itself or it's something that we're all getting but not noticed (and i would notice this!)

The fact that it doesn't do it in all the ISOs is unusual as well.

Are you able to swap the body? Or borrow one to check?

cheers
Paul

 

Heya Paul,

I'm not able to swap the body for another one (I'm the only person I know with an fp in my circle of photography friends) but I could certainly try to see if this happens at 10bit and 8bit, as I've as yet only used 12bit for capture. 

Here you go Paul (or anyone who's curious):

https://drive.google.com/open?id=1tgJ7nNLGzTOXLvVRJkn06yr8PGBgRNcF

Folder for ISO 320 with 24 frames, and a folder for ISO 800 with 24 frames. The flicker at 800 is pretty crazy so hopefully that's a treasure of information for you to dive into. 

1 hour ago, paulinventome said:

Did you post a couple of DNGs before? I think you did?

What would be interesting is that in the DNG there is a value called Brightness Value which i noticed changes with ISO. If this is being changed in the DNG by mistake then it may explain the data. Have you still got the same DNGs handy?

cheers
Paul

I don't have those original ones from pages back, but I can post some frames from 320 (which doesn't flicker) and some from a flickering ISO if that would be useful!
Edit: Uploading a 1 second DNG sequence from both 320 and 800 to google drive. It'll be ready in a moment. 

24 minutes ago, Noli said:

My flicker is still there as well, I am only using dumb lenses so thats not the issue.

Was really hoping they would fix this and the audio timecode not being embedded in the wav correctly.

Looks like we're a part of a lucky few, Noli  What's weird to me is that the flicker seems to affect the red and blue channels more than green. 

Wouldn't that indicate something happening with white balance? It makes no difference having a WB preset selected like Sunlight, or manually dialing in 5500K instead. Tried it with disabling all of the auto lens corrections, with autofocus, manual focus etc. I feel like I've tried most variables that would affect it. 

32 minutes ago, paulinventome said:

So between 320 and 3200 it flickers still? Which is the worst and what lens?

cheers
Paul

Heya Paul, I'd say the flickering is pretty consistent between ISOs where it occurs. It would be tough to say one is worse than another. 

This is with the 45mm Sigma lens, but it happens the same on the Sigma 35mm. I do have a Pentax K to L-mount adapter where I could test some "dumb" lenses to rule out the lenses, however. 

I did also try out adding the Deflicker node in Resolve, and that seems to make it 99.99% visually undetectable, which is nice. But the thought of using software to fix things after the fact makes me lose sleep at night

https://imgur.com/2c9IcMu
(In case the google drive video is acting up)

Just installed 1.02, and unfortunately it seems like my black level flickering issue is alive and well at the same ISOs where it was before. On the low end, the highest I can go without flickering is 320, and on the high end, the lowest I can do without flickering is 3200. Granted, the image is clean at both of those ISOs, but at this point I may need to send in the body so they can figure out what happened to my unit since this doesn't seem to be a widespread issue. 

Edit: Here's what it looks like in Resolve during playback (4k screen capture). Notice the jumping black levels in the waveform. 
https://drive.google.com/open?id=1kMpU9QFj92Ly-h7aOkN5Asb9ombpMolX

The sole reason I joined Twitter a few days ago has paid off. Mr. Yamaki replied to my question asking if lens aberration correction would apply to cDNG and .Mov:

Question about the 1.02 firmware notes.

"...and introduces added lens aberration correction processing."

Does this mean the cDNG raw files will have an option for lens corrections like distortion and vignette removal burned in, or would this just be for .MOVs? I bought the fp for its raw capability, so I don't imagine ever recording to a video format. It's fun being able to treat video like motion stills

On the other hand, programs like Resolve don't support lens correction databases or metadata that I know of like Camera Raw does. It would be great if they did!

Just now, Chris Whitten said:

Are you using the body with a Sigma lens, or in some kind of auto exposure mode?

All my frames are exactly the same exposure, but I don't have an L series lens, so I'm using old manual like Leica M or Takumar.

I have found those annoying bugs in the stills function of the camera, but so far the cine mode has been bug free for me.

Both Sigma lenses: 35mm Art and the 45mm DG DN  Fully manual mode for cine. 

Definitely one of the strangest cameras I've yet owned, haha. 

I've seen quite a few initial frames that are exposed about a stop hotter than what I've dialed in for a video, and this is on 1.01. 

On a call with Sigma support the tech wondered if the firmware didn't properly update from the factory somehow, since I received my camera with 1.01 pre-installed. After numerous re-installs and resets, the behaviors with exposure and flickering haven't changed. If 1.02 doesn't fix these issues, I'll send the body to Sigma NY to let them diagnose what's wrong and nuke the software if need be

You know what would be handy? A sensor whitepaper written by Sigma. Just a blast of all the tech details about the sensor they're using so that we don't have to be quite so investigative. Surely that wouldn't be a big deal for the engineering team to put together. 

5 hours ago, Chris Whitten said:

CrimsonEngine on Youtube has tested the FP video quite extensively, all the CDNG settings for example, also looked at grading in Resolve.

There is no stated native ISO from Sigma as there is for the Blackmagic Pocket, so I asked him what he had found was closest to 'default' ISO in his tests and he replied ISO640.

In any case, my music space is very low light and I live in the UK (also low light), so around 640 was the minimum ISO I could use without using artificial light.

The fp has some interesting ISO behaviors, certainly!

I've been using it at lower ISOs even in the dark and then pushing things 2.7 stops without major issues. It's truly a form of night vision. Shooting wide open at F1.4 helps as well, but consider this example of a quick night shot I did:

First shot is at ISO 320 pushed 2.7 stops, second shot is ISO 2500 as-is, third shot is ISO 320 pushed 2.7 stops again, and the last shot is ISO 2500 as-is.

The original video file straight out of Resolve can be found here to avoid seeing the recompression macro blocking artifacts from YouTube:
https://www.youtube.com/redirect?q=https%3A%2F%2Fdrive.google.com%2Fopen%3Fid%3D1sEtAlhfZOjfHO7R7XmXrJJvMUiJqRAHC&event=video_description&v=vYpPZPCWo5g&redir_token=-6uwx0ZKXpomIasN7CDyU0Ps-m98MTU4NDAzNTQzNkAxNTgzOTQ5MDM2

1 hour ago, Lars Steenhoff said:

Make a report to sigma, if more people who do this it will get higher on the priority list

Heya Lars, is there a particular way that's best to make a report to Sigma? I've just been using their email address. Though I don't often get a reply back. 

I think I've experienced the exposure bug with the stills mode. Here's what happened:

Yesterday, I had the camera in Stills mode in Manual. I took a reading with my incident meter and the settings were F2.8 at 1/160th. I dialed those settings in, then took a photo. Image appeared dark. Tried it again, same result.

Then I checked those images with RawViewer, and it turns out instead of 1/160th, the exposure was actually 1/640th, or two full stops down from where I told it to be. 

After resetting the camera back to default settings then going back into Manual mode, this issue disappeared.

What in the world?

A camera that doesn't use the settings I dialed in for a photo is not a reliable camera 😛 

3 minutes ago, rawshooter said:

"SIGMA is also developing firmware Ver.1.02, which corrects card errors when using certain SD cards, and introduces added lens aberration correction processing. This is scheduled to be released on Wednesday 18th March 2020."

I wonder whether this will fix the manual exposure bug. Doesn't sound like it...

I'm hoping that post is a case of focusing on the major new feature without listing all the nitty gritty of what else is included!

https://www.sigma-global.com/en/news/2020/03/02/2581/

Official announcement

6 hours ago, Lars Steenhoff said:

it would look the same except one difference is 12 bit for the video vs 14 bit for the still

Makes sense! Not that I see any huge need for 6K as-is for the sake of it any time soon, other than having more padding for compositional adjustments or post-stabilization in the vein of what Fincher does. I think even with Gone Girl they captured at 6K, cropped and stabilized for 5K, then delivered a 4K downsampled master from that. Resolution is definitely the one thing we have plenty of from most cameras.

5 hours ago, rawshooter said:

No, because it's RAW sensor data - all sampling is done in post (except for the downsampling from 6K to 4K).

Stupid question: If it were possible to see the 6K before downsampling, would it look any different than a photo in still mode? 

48 minutes ago, Chris Whitten said:

I'm not sure about sales. Several people have been dumping the FP on Ebay UK presumably out of frustration.

There is virtually nothing on the web about stills photography using the FP except the first few blog posts that are now months old. 

Very little on Youtube except people uploading 1 minute test footage clips without much info. I was an early original BM Pocket user and there was much more dialogue and information sharing going on then. Thank heaven for this discussion, also one or two Youtubers who are testing and discussing.

I wonder if it's a case where people aren't so keen on Sigma's new entry into the cinema camera market since people associate them with slow but gorgeous "weird" cameras? Almost like if BMW were to suddenly decide to make a cell phone

Every modern camera seems to be a great sensor in a box with the differentiation being software at this point, save for the obvious technical advantages of the higher-end cameras. If Sigma can make the fp platform stable, reliable, and consistent, I don't see why they couldn't continue the line well into the future. 

Nice find, Lars!

https://www.pronews.jp/news/20200210142806.html

Page should auto-translate from Japanese in Chrome.