Attila Bakos

Does anyone have a recommendation for SD cards that work with 400Mbps in the X-T3? The cards recommended by Fuji are an overkill as far as required bitrate goes, and they are very expensive as well. Fuji says that V60 cards should work fine for 400Mbps, but I remember that this is not as easy as it sounds, some V60 cards had problems in the GH5, while some worked perfectly fine.

I'll look into this as well when my X-T3 arrives. Ordered it today.

@CyclingBen

I didn't mean to investigate, but googling X-H1 update 1.10 has your thread in the top 5 results

I thought the auto ISO problem was fixed so that the transitions are really smooth now. If it's not, then the problem is not back, but it's still here... I just thought it is not as visible as in the first video I posted.

10 minutes ago, CyclingBen said:

They did fix it and it wasn’t iso shifting that was the problem, it was an exposure compensation shift even if exposure comp was turned off.

In the linked video, that is auto ISO being used. If you set the ISO to a specific value you won’t see that happen.

Nope it was an auto ISO problem, Zed even mentions it in this video from 6:40:

Edit: funny thing is that you as the creator of this thread specifically mention smooth auto iso as a feature of v1.10:

26 minutes ago, Mattias Burling said:

ISO doesn't change unless you accidentally left it in auto. I wonder if he did that or maybe even left the DR setting in auto? I've never tested either settings so I don't know what it would look like.

My point is that having these bad ISO transitions looks to be a step back, since this problem was fixed via firmware 1.10 on the X-H1.
It would be great if someone from here tested this though.

It seems that the ugly ISO changes are back with the X-T3, check from 0:16. Can someone confirm this?

Does anyone know a smart way to mount this to the X-T3? Anker PowerCore+ 26800 PD

According to Fuji, it should give you 400 mins of 4K60 recording time, compared to 40 mins with the battery.

17 hours ago, androidlad said:

As discussed before, HEVC smoothing filters SAO and SIS contributed to that waxiness.

Out of curiosity I created two videos from a UHD sized TIF with lots of fine detail, once with SAO and SIS enabled, then disabled. I used the x265 library included in ffmpeg using CRF 3 for a raised bitrate and no other parameters besides SAO/SIS. I see absolutely no difference between the two versions even at 400% magnification. Then I thought maybe it's because there's no movement as it's a still image. So I downloaded some DNG's from Blackmagic's website, exported a short clip in Resolve using lossless AVI, and used the same ffmpeg scripts. Again, zero difference between the smoothed and non-smoothed versions. I checked the files with mediainfo to see the if the encoding parameters were passed to the x265 library and they are there.

To make it short: if the hardware encoder works in a similar way (and I didn't fuck up my test somehow), I wouldn't worry about detail loss.

6 minutes ago, androidlad said:

As discussed before, HEVC smoothing filters SAO and SIS contributed to that waxiness.

Could you in theory compare it to 200Mbps 8bit H.264 to see if it really makes a difference? I mean, does the H.264 recorded by the X-T3 have similar smoothing filters?

32 minutes ago, androidlad said:

SAO and SIS are essentially smoothing filters and are bitrate independent. UHD BD authoring tools have these turned off.

http://iphome.hhi.de/wiegand/assets/pdfs/2012_12_IEEE-HEVC-Performance.pdf

Well, all we need is someone with an X-T3 and an external recorder to see if there's any visible difference.

30 minutes ago, androidlad said:

Yes, these two filters are designed to preserve overall image quality at low bitrate, at 200/400Mbps bitrate Fuji implemented, they really should disable those filters to allow more fine detail to come through.

But do they really reduce fine detail at these high bitrates? Would love to see a comparison, maybe this is for pixel peepers only.

4 hours ago, androidlad said:

X-T3 10bit encoding anaylsis from Weibo:

https://www.weibo.com/KazuoHiraiSONY?is_hot=1#1538181717102

X-T3 10bit HEVC encoder has SAO (Sample Adaptive Offset) and SIS (Strong Intra Smoothing) filters enabled, which unfortunately reduces high frequency detail. 

While that might be true, both are designed to reduce artifacts like banding, blocking, ringing, and both are enabled in x265 by default, so I guess it's nothing to worry about.

Would you guys consider the XF16-55 over the XF18-55 for video? Does the IQ difference translate into video as well? Is the XF16-55 smoother in video AF? I know the big guy doesn't have OIS, but I'm on tripod/gimbal most of the time.

4 minutes ago, Ingerson said:

I used DR400. 

Thanks, I thought the difference would be smaller.

2 hours ago, Ingerson said:

nonscientific DR comparison between Eterna simulation and F-Log. Mid exposure, then one and two stops up and down. Included waveform.

https://www.dropbox.com/sh/rw3ecm3tnft1a21/AADf4nRWTJ8WqSQK_TWKiKXUa?dl=0

No idea how to measure that into numbers, sorry.

Eterna is in DR100 mode right? In DR400 mode the dynamic range should be about the same, according to Fuji.

4 minutes ago, jonpais said:

I seem to recall that this X-T2 video was shot with the 16-55 f/2.8, but it's been a while, so I'd have to research it again.

It's on vimeo as well, the description says 18-55: https://vimeo.com/195452189

It would be nice if Fujifilm gave us a hint about X-H2 release date. Can you please ask them about it?

19 minutes ago, androidlad said:

The newer f/2 primes and f/2.8 zooms are quad-linear motor so AF-C performance is much improved. On X-T3 older gen lenses got a bump in performance but still cannot compete with the newer ones.

Only the lenses with LM in their name use linear motor. The newer f/2 primes (23,35,50) use some other technique. And I believe only the XF90 is quad linear. The XF 50-140 is triple linear, and the XF 16-55 is twin linear.

The Fringer adapter is looking good on the X-T3:

AF test with older primes compared to X-T2. X-T3 focuses faster with less hunting, also continuous AF seems to be more confident, however the transition between close and far subjects is not butter smooth. Might be better with newer lenses.

25 minutes ago, toxotis70 said:

is there any comparison between 420 and 422 , with any other camera ?

I didn't find any so I decided to do my own test. I downloaded two freely available Fuji RAF files and exported a 3840x2160 area as 16bit TIFF. Then I used these files to create 1 sec movies using FFMPEG. I created 3 versions of each: 8bit 4:2:0, 8bit 4:2:2, 10bit 4:2:0. I used H.264 encoding, the files should be playable in the latest Resolve/Premiere. All are 200Mbps. You can check the files in MediaInfo to verify encoding parameters.

Download the test from here: https://drive.google.com/open?id=1SbeLZJTN7jWLC2XStb-EU4QWIsCQRDyD

For some reason the TIFF has slightly different colors, ignore that. What matters is the transition of colors. Try to pull keys in the first test, or break the sky gradient in the second test with a high contrast. Tell me if you see a huge difference, because I don't. Of course my test can be flawed, I'm no expert in encoding.

12 hours ago, Luke Mason said:

Ha, I've done my research, The bitrate Fuji implements at 200Mbps and 400Mbps exceeds the maximum for hardware acceleration (for now).

But then how is it possible that my old i7-3770 with a GTX 1080 plays the 400Mbps jellyfish sample smoothly?

I downloaded the sample from here: http://jell.yfish.us/

I used MPC-HC to play the files. 10% GPU usage and around 15% CPU. Do NLE's handle these files differently?

11 minutes ago, Jimmy said:

Care to post a link to these tests?

The XT2 has at least 12 stops... good enough for most. XT3 should improve this further.

I'd be interested too as both Cinema5D's and Dpreview's initial comparison shows slightly more DR on the X-T3.

The lockup issues are still here though. Just saw a video where the guy had three lockups in a day with Fujifilm recommended SD cards. Only battery pull helped. The camera was the final version. He said he will write an article with details. The video only shows the camera in a locked up state so I'm not linking it.