Fuji X-T3 and X-T4 discussion

[thephoenix]

39 minutes ago, Anaconda_ said:

Here's a screen grab of the images in Resolve, the left screen is the XT3 file and the right is from the Ninja. I set them both to 300 and lined them up to form basically one image with a line in the middle. I thought that would be the easiest way to see if there's any difference in the colour. - both are set to Full in Clip Attributes. To my eyes, they're exactly the same, but if there is a difference, I wouldn't even think about it while shooting.

One thing I did notice though, is the capture from the Atomos is 2 or 3 frames behind the file straight from the camera, nothing major, and I'm sure it's not camera / recorder specific, but worth noting.

 

Edit:

If people want, I could also post screen grabs from the colour page, showing the histogram, vector scope, curves etc. But they don't change at all between the clips. Safe to say they're exactly the same image from that perspective too... 

Weird i have a difference when transcoding the file. 

[Anaconda_]

12 minutes ago, thephoenix said:

Weird i have a difference when transcoding the file. 

I didn't transcode anything, this is the OG files directly into Resolve. 

[thephoenix]

Just now, Anaconda_ said:

I didn't transcode anything, this is the OG files directly into Resolve. 

I know i was talking about my test  on the previous page

[BrunoCH]

3 hours ago, Attila Bakos said:

For what exactly do we need more accuracy? We know all about F-Log, all the info is there in the data sheet provided by Fujifilm.

In Resolve I want to use RCM without LUT. Now I have to choose for the input colorspace REC2020 scene (gamma around 1.9) or  REC2020 gamma 2.4. Both are correct  for gammut but the two gammas are neither completely wrong nor completely right. We know the conversion formula in Fuji data sheet but I can’t write this formula in Resolve. Same thing for the data levels. But I think it's more easy to compensate for not very accurate data levels than to correct the gamma.

[Llaasseerr]

7 hours ago, Anaconda_ said:

Here's a screen grab of the images in Resolve, the left screen is the XT3 file and the right is from the Ninja. I set them both to 300 and lined them up to form basically one image with a line in the middle. I thought that would be the easiest way to see if there's any difference in the colour. - both are set to Full in Clip Attributes. To my eyes, they're exactly the same, but if there is a difference, I wouldn't even think about it while shooting.

One thing I did notice though, is the capture from the Atomos is 2 or 3 frames behind the file straight from the camera, nothing major, and I'm sure it's not camera / recorder specific, but worth noting.

 

Edit:

If people want, I could also post screen grabs from the colour page, showing the histogram, vector scope, curves etc. But they don't change at all between the clips. Safe to say they're exactly the same image from that perspective too... 

I've tried this out in Resolve 16b3 with the same files and unlike Resolve 15, I need to set the ProRes clip to Video levels, not Full levels. Besides that, the colours now apparently match with the latest bug fixes. However I'm getting a separate luma clipping issue. Details over here in the original thread.

[thephoenix]

10 hours ago, BrunoCH said:

In Resolve I want to use RCM without LUT. Now I have to choose for the input colorspace REC2020 scene (gamma around 1.9) or  REC2020 gamma 2.4. Both are correct  for gammut but the two gammas are neither completely wrong nor completely right. We know the conversion formula in Fuji data sheet but I can’t write this formula in Resolve. Same thing for the data levels. But I think it's more easy to compensate for not very accurate data levels than to correct the gamma.

same here.

for flog i am loading the fuji flog to rec709 lut in rcm so rcm should still be working.

[andrew_dotdot]

Umm. I need a TSDU version, I think. (Too Stupid Didn't Understand) Did something change in the new Resolve 16b3?

For me the use case is F-LOG on internal storage with h.265 in Resolve, which is I guess will be the use case for the majority of us interested parties in the dum-dum crowd.

 

[Lars Steenhoff]

1 minute ago, andrew_dotdot said:

Umm. I need a TSDU version, I think. (Too Stupid Didn't Understand) Did something change in the new Resolve 16b3?

For me the use case is F-LOG on internal storage with h.265 in Resolve, which is I guess will be the use case for the majority of us interested parties in the dum-dum crowd.

 

yes something did change in the latest resolve b3, check the release notes and it mentions fuji files

[Anaconda_]

8 hours ago, Llaasseerr said:

I've tried this out in Resolve 16b3 with the same files and unlike Resolve 15, I need to set the ProRes clip to Video levels, not Full levels. Besides that, the colours now apparently match with the latest bug fixes. However I'm getting a separate luma clipping issue. Details over here in the original thread.

Just to clarify, my test was also in Resolve 16b3 with both files set to Full levels, and they matched according to the charts in the colour tab.

[andrew_dotdot]

2 hours ago, Lars Steenhoff said:

yes something did change in the latest resolve b3, check the release notes and it mentions fuji files

Yes, I know. I'm the overwhelmed OP for the question -- I innocently asked if anyone knew what the change referred to in the release notes for 16b3 specifically meant for the X-T3. ?

 

1 hour ago, Anaconda_ said:

Just to clarify, my test was also in Resolve 16b3 with both files set to Full levels, and they matched according to the charts in the colour tab.

Does this mean...

1. The erroneous color shift for the internally-recorded h.265 files in Resolve is now solved? I.e, it matches what you'd get from an external recorder? (caveat: h.265 set to "Full Levels" and ProRes set to "Video Levels) 

2. The new 16b3 has introduced a situation where you must manually set ProRes F-Log clips to "Video Levels" (not Auto) in the Clip Attributes

3. A bug has been introduced in 16b3 causing a problem with highlight clipping for h.265 internal clips. (no inside-Resolve workaround found as yet, according to the thread with @Llaasseerr)

[Anaconda_]

16 minutes ago, andrew_dotdot said:

1. The erroneous color shift for the internally-recorded h.265 files in Resolve is now solved? I.e, it matches what you'd get from an external recorder? (caveat: h.265 set to "Full Levels" and ProRes set to "Video Levels) 

2. The new 16b3 has introduced a situation where you must manually set ProRes F-Log clips to "Video Levels" (not Auto) in the Clip Attributes

I'm not sure why this keeps being said. I set BOTH clips (h265 AND ProRes) to FULL LEVELS and they matched in 16b3.

[andrew_dotdot]

47 minutes ago, Anaconda_ said:

I'm not sure why this keeps being said. I set BOTH clips (h265 AND ProRes) to FULL LEVELS and they matched in 16b3.

Perfect! Good that you clear this up. (Aw, nuts! I wanted to change my post above, but it was too old now.)

[thephoenix]

speaking of ninja, got my cage today, pretty cool

[BrunoCH]

15 hours ago, thephoenix said:

same here.

for flog i am loading the fuji flog to rec709 lut in rcm so rcm should still be working.

No it shouldn't for input. Still works for output. RCM works without LUT. If you load a LUT in project setting, it's something like you add, automatically, on all your clips, a first hide node with this LUT. 

[Llaasseerr]

11 hours ago, Anaconda_ said:

Just to clarify, my test was also in Resolve 16b3 with both files set to Full levels, and they matched according to the charts in the colour tab.

Thanks for letting me know that you are getting consistent behaviour. I think there's some issue just on my Mac, where Resolve full range is not matching h.265 full range because when I rolled back to Resolve 15, I suddenly had the same issue where previously 15 had behaved the same as what you're saying. I have no idea what has happened on my computer because I haven't updated the system software. I've reported this to BMD but they have no idea either.

[thephoenix]

7 hours ago, BrunoCH said:

No it shouldn't for input. Still works for output. RCM works without LUT. If you load a LUT in project setting, it's something like you add, automatically, on all your clips, a first hide node with this LUT. 

ok, but i never found a clear workflow for rcm. and the more i read the more i get confused.

 

7 hours ago, BrunoCH said:

No it shouldn't for input. Still works for output. RCM works without LUT. If you load a LUT in project setting, it's something like you add, automatically, on all your clips, a first hide node with this LUT. 

to me the real question is "will it still be there after transcoding to prores or dnxhr" because you don't want to work with h265 files directly in resolve. or you have to create proxy files.

did a test and i had a clear difference in red

[Inazuma]

Can someone tell me how good autofocus is in video with EF speed boosted lenses?

[andrew_dotdot]

31 minutes ago, Inazuma said:

Can someone tell me how good autofocus is in video with EF speed boosted lenses?

I’ve got the Fringer EF-X Pro1 and I can tell you it’s not great for video. I only do video with the X-T3. It can take a second or two to acquire focus. Then, for thin DOF shots in “C” mode, it will “micro-hunt” quite a lot while shooting.  I have a button assigned to focus-lock to kill that during a shot. I have used it with the 24-105 f4L and the 70-200 f2.8L, and a bit with the 50mm f1.8 STM lenses that I already had, and wasn’t satisfied. As salt in the wound, it ignores the focus sensitivity and speed settings in the camera. It will shift focus as fast as it can, no matter what the camera-menu settings.

I now use the Fuji 18-55mm and 10-20mm stabilized lenses for video. They’re good lenses, and I’ve been satisfied with the AF performance in run-and-gun situations, whereas I’ve pulled my hair out using the Fringer adapter and my wonderful Canon glass.

[thephoenix]

3 hours ago, Inazuma said:

Can someone tell me how good autofocus is in video with EF speed boosted lenses?

depends on lense af and motors. it is pretty good on my tamron 24-70 and my 28mm 2.8. i mean when i compare to the same lenses on a canon body.

i also have canon 85mm 1.4 but didn't tried it yet on the fuji

to me the problem is not the adapter but the lense's af itself.

[andrew_dotdot]

@thephoenix >> Just curious - what Canon body do you have?  Mine was a 70D for any auto-focus-style work (which I sold to fund the X-T3). I was really surprised to read your experience, since mine was just the opposite — same lenses had wildly different performance on the Canon vs the Fuji body for me.

2 hours ago, thephoenix said:

depends on lense af and motors. it is pretty good on my tamron 24-70 and my 28mm 2.8. i mean when i compare to the same lenses on a canon body.

i also have canon 85mm 1.4 but didn't tried it yet on the fuji

to me the problem is not the adapter but the lense's af itself.