Fuji X-T3 released

[Grug]

I asked the tech guy and he replied "I'm not sure if there hardware limitations to get the PIX to 60fps, but I'll add that to our Feature Request list as well."

I agree the LR unit is very convenient. However, I sense that it wasn't a big seller.

Interesting. I love Sound Devices’ monitoring features and robustness, but I always thought the LR unit was the primary appeal they offer - since when you’re recording your video externally, it’s usually from cameras with limited audio features.

I wonder if they’re either on the cusp of releasing something new and exciting (in the same space as the PIX/LR module), or whether that lovely, unified external video/audio recording concept is about to be lost to us forever?

 

[NorBro]

Look, man...

I hear you. Loud and clear.

The disconnect is truly a technicality.

Hear me out...

_____

I am simply comparing the film stock simulations to the others' offerings by choosing to compare apples to apples (IN MY OPINION). It doesn't matter if Eterna is not voted as the best choice to be compared to neutral or natural. It can be any of the other 6 choices.

However, since Provia is called 'Provia/Standard' then I think it's fair to say that that one would be compared to Canon's/Panasonic's/Sony's/Nikon's 'Standard'. Since Velvia is 'Velvia/Vivid' then I think it's fair to say that that one would be compared to C/P/S/N 'Vivid'.

Now we have 5 left; Astia, Classic Chrome, PNH, PNS and Eterna.

Now with the other cameras, you may find landscape, portrait, faithful and neutral/natural. Sony and Panasonic also have their Cine gammas. Then there is HLG in a few. And then the LOGs; everyone has their own LOG(s).

_____

Friend. Do you honestly believe that these Fuji film simulations are anything different and something special?

Come on, man...I know you've been around the block for a long time. Through the Blackmagics, through the Canons.

They are picture profiles/styles with subtle differences that corporate is providing us. They have been there as options for video the second Fuji started providing video in their stills cameras, and have looked very similar many years ago:

https://www.youtube.com/watch?v=GNxeybiiXoo (random sample)

It's brilliant to name them after film stocks because the company makes film!

I don't care what this digital Eterna is according to Fuji. It can be 250D 8663, 500T 8673, 500T 8573; it is completely irrelevant in my main argument. I'm sure they did an amazing job and provided some perfectly accurate digital Eterna stock in this digital camera to the best of their superior ability, but the X-T3 video with the Eterna film simulation does not look like film. It is a picture profile/style that tries to SIMULATE a look, a stock.

_____

All I am saying is that I choose Eterna or Pro Negative Standard as my comparison to neutral/natural on C/P/S/N cameras. It's not going to be a perfect comparison, of course, especially since Fuji cameras are pretty punchy, but that's just my personal choice.

IMO, when someone is asking if the camera is capable of a more neutral look then the answer should point to one of the 6 film simulations, not F-Log and post-production. Because in post you can almost do anything. You can manipulate the heck out of any of them and call it "neutral".

If you shoot F-Log and then add a LUT in post and call that neutral (and of course it very well could be and is) then that in my opinion is in a different category and doesn't count.

You wouldn't hand over LOG footage (as you mentioned) to people who asked for neutral footage unless they specifically asked you for LOG footage.

It doesn't count...and this is our disconnect.

A technicality, IMO.

I mean, am I crazy? Am I not providing a valid argument?

 

[Joshua_G]

…
…
I mean, am I crazy? Am I not providing a valid argument?

You aren’t crazy in my eyes. The issue became an argument, or a debate, mostly due to semantics nuances.
Happy New Year to all and happy shooting.

 

[David_Manning]

Just jumped onto the XT-3 bandwagon. I’m blown away by the 400mbits 4k 10-bit 4:2:0 footage...but I’m impressed even more by the autofocus capabilities during video capture. I never thought I’d see a Super35 sensor auto-focus that well.

 

[puredrifting]

Just jumped onto the XT-3 bandwagon. I’m blown away by the 400mbits 4k 10-bit 4:2:0 footage...but I’m impressed even more by the autofocus capabilities during video capture. I never thought I’d see a Super35 sensor auto-focus that well.

Welcome to the club! I'm a new member as well, it is a paradigm changing camera too as far as I have experienced. The dark secret is, I actually overall like the images better than the ones coming out of my C200 most of the time from a camera that cost 1/5th as much.

 

[Jim Feeley]

Just jumped onto the XT-3 bandwagon. I’m blown away by the 400mbits 4k 10-bit 4:2:0 footage...but I’m impressed even more by the autofocus capabilities during video capture. I never thought I’d see a Super35 sensor auto-focus that well.

So I'm looking at the camera, too. David, Dan, and everyone: This has probably already been discussed (and missed by me), but how would you compare the X-T3's AF to Canon DPAF? Especially in handheld verite(ish) situations? Not just in terms of better/worse, but how do they differ?

 

[macgregor]

Whaaaaaaaat?

 

[jcs]

So I'm looking at the camera, too. David, Dan, and everyone: This has probably already been discussed (and missed by me), but how would you compare the XT-3's AF to Canon DPAF? Especially in handheld verite(ish) situations? Not just in terms of better/worse, but how do they differ?

Full sensor DPAF- as good or better than Canon for video; see differences here:

Sony with MC11 seems to work ok for video with Canon lenses. Any options for usable Canon lens AF-C and the XT3 (Fringer?)?

 

[Greg_E]

When comparing AF performance you really need to specify still or video. Stills, the Canon and Nikon (not cinema/video cameras) are still better, but the margin decreases every generation. I'll not comment on the video because I haven't experienced it, rumors are that it is getting pretty good during video.

 

[marvinhello]

So I'm looking at the camera, too. David, Dan, and everyone: This has probably already been discussed (and missed by me), but how would you compare the XT-3's AF to Canon DPAF? Especially in handheld verite(ish) situations? Not just in terms of better/worse, but how do they differ?

Full sensor DPAF- as good or better than Canon for video; see differences here:

Sony with MC11 seems to work ok for video with Canon lenses. Any options for usable Canon lens AF-C and the XT3 (Fringer?)?

Overall Canon's DPAF (Dual Pixel AF) is better than X-T3's PDAF (Phase Detection AF). It's more accurate tracking, less hunting and more consistent across different lighting scenarios.

One of the reason being, Canon's DPAF is built into the pixels (split in half), it's 80% of the actual photosites, on 1DX II, it has around 17MP AF pixels. On X-T3 the PDAF pixels are masks and "only" cover 2.16MP across the sensor.

Canon's DPAF also offers more tracking modes, while X-T3 only do full auto (no user control), area (focus box) or face tracking, and face tracking becomes extremely unreliable when the face becomes small in the frame.

 

[jcs]

Overall Canon's DPAF (Dual Pixel AF) is better than X-T3's PDAF (Phase Detection AF). It's more accurate tracking, less hunting and more consistent across different lighting scenarios.

One of the reason being, Canon's DPAF is built into the pixels (split in half), it's 80% of the actual photosites, on 1DX II, it has around 17MP AF pixels. On X-T3 the PDAF pixels are masks and "only" cover 2.16MP across the sensor.

Canon's DPAF also offers more tracking modes, while X-T3 only do full auto (no user control), area (focus box) or face tracking, and face tracking becomes extremely unreliable when the face becomes small in the frame.

I have a little experience with Canon DPAF & Canon color (just sold my C300 II, primary camera (video & stills) is the 1DX II, and small/travel cam is 80D). Canon DPAF is pretty good though not perfect. The Sony A7 III looked to have achieved near parity with Canon DPAF for video AF (from studying many examples on YouTube).

After studying video examples online, Fuji has hit a home run with the XT3. I only owned a single Fuji camera (Real 3D W3) and while highly innovative, wasn't very useful. The XT3 has better color (more accurate) than Canon or Sony out of the camera (I always use highly modified picture styles for Canon and Sony, never use the stock settings), and from what I've seen so far, has AF-C for video that's on par with anything from Canon or Sony. Do you have examples (YT etc.) that show Canon/Sony outperforming the XT3 in real world shots for AF-C + video?

Fringer + XT3 + Canon 600mm: seems to work pretty well (end result looks stable/usable):

Sony's A7 III was groundbreaking given the quality provide for $2k. Fuji's XT3 for $1400 + the availability of a decent Canon adaptor + really nice native lenses is impressive.

All face autofocus algorithms know basically where the eyes are, being able to select/show what going on (and/or matching the eye vs. estimation), especially for video is useful. Sony provides this feature as well.

Perhaps the XT3 will push Sony to release the A7S III with much improved color science (VENICE etc.) and perhaps less crippling, e.g. at 10-bit option (internally), though that's not a deal breaker (1DX II 422 8-bit works fine when doing 'color in camera'). Probably going to sell the 80D and give the XT3 a spin.

EDIT: XT3 vs. EOS-R skintones and AF tracking:

Looks like Canon no longer has the lead with color; perhaps less gimping on resolution/bit-depth/codecs/features for the future.

XT3 rolling shutter looks OK for the value proposition:

Another color example for skin tones (hardest task to do right with color): Fuji doing better than Canon (and much better than Sony):

For stills, my models/clients are extremely picky when it comes to skin tones. In the tests I showed them for video clips, they always preferred Canon color to Panasonic and especially Sony. So I never considered using anything but Canon for stills (Nikon would have been an option but don't own any Nikon lenses). From what I'm seeing so far in brief research, the XT-3 looks really good for skin tones, better than Canon for video. I shoot raw stills and process in Photoshop- from prior posts has the 'worming' issue been figured out/resolved?

If there was a decent focal reducer which provided usable AF-C video performance (to match 1DX II FF Canon lens performance in video (actually better: re: 10-bit 420 (detailed) 4K > 8-bit 422 (soft) 1080p))), that would make the XT-3 an even greater bargain. As it stands it looks like a very good upgrade from an 80D or similar camera (travel / mixed use camera).

 

[yoclay]

There is something I have noticed with the skin on Fuji. It looks less blotchy than on a Sony or a Canon. A bit flatter, smoother, slightly warmer. In a similar way that a Cooke lens renders skin in fact. However, I do wonder if some kind of smoothing or treatment is going on in camera. Perhaps some kind of noise reduction algorithm ? I like it, but it is a look. Anyone else seeing what I am seeing ?


If you look at Macgregor's video with an X-T2, you will understand what I am speaking about. As a colorist I am very aware of the texture coming from different cams, and it something I am noticing consistently with these little Fuji's..


https://www.youtube.com/watch?v=WmeOwkvg35g

 

[jcs]

There is something I have noticed with the skin on Fuji. It looks less blotchy than on a Sony or a Canon. A bit flatter, smoother, slightly warmer. In a similar way that a Cooke lens renders skin in fact. However, I do wonder if some kind of smoothing or treatment is going on in camera. Perhaps some kind of noise reduction algorithm ? I like it, but it is a look. Anyone else seeing what I am seeing ?


If you look at Macgregor's video with an X-T2, you will understand what I am speaking about. As a colorist I am very aware of the texture coming from different cams, and it something I am noticing consistently with these little Fuji's..


https://www.youtube.com/watch?v=WmeOwkvg35g

Back when that video was posted, YT 1080p was somewhat degraded quality, vs. upscaling a 1080p video (+ slight sharpen) to 4K and uploading as 4K (and watching in > 1080p on YT). Lately it appears (in Chrome) YT 1080p is much improved. The point being that it's hard to see critical detail on older 1080p YT videos. It's clear that MacGregor tuned the color on those cameras. Skin tones look decent on both, though when you look at white and gray objects, you can see shifts to green/magenta, even the sky looks different. Overall Fuji looks more realistic. XT-3 has even better color?

As for face-specific processing: e.g. performing extra NR and/or lower contrast on face to reduce wrinkles and imperfections: it wouldn't be a bad option as long as it can be turned off.

In the video I posted with EOS-R, A7 III, XT-3 using basically auto everything, the Fuji looks the most accurate/realistic. Canon is classically too red/yellow (which is R+G) and Sony is way too yellow (R+G): neither is accurate. Fuji could use a hair less magenta, but it's decent as is (better than a green bias). That being said, those of us with the experience and inclination can tweak in-camera color and get much better results (as well as post treatment). What I really liked about the 5D Mark II and III was great color out of the camera (a bit too red but usable). Canon went backwards in color after the 5D3. Even with extensive tuning, I couldn't get the C300 II to match the tuned skin tones on the 1DX II. At this point I really don't want to spend time color tuning a new camera: it should be accurate using default settings (which is why ARRI is still #1 in the pro space).

Major kudos to Fuji with the XT-3- leap-frogging not only Panasonic and Sony, but also Canon for less money! Since Fuji makes a MF body, curious if they'll ever make a FF body: very useful for stills regarding ease of super shallow DOF (FF lenses still provide most affordable options; already have many FF lenses); video it's cool but not as important (don't shoot ultra shallow DOF video very often).

 

[yoclay]

Back when that video was posted, YT 1080p was somewhat degraded quality, vs. upscaling a 1080p video (+ slight sharpen) to 4K and uploading as 4K (and watching in > 1080p on YT). Lately it appears (in Chrome) YT 1080p is much improved. The point being that it's hard to see critical detail on older 1080p YT videos. It's clear that MacGregor tuned the color on those cameras. Skin tones look decent on both, though when you look at white and gray objects, you can see shifts to green/magenta, even the sky looks different. Overall Fuji looks more realistic. XT-3 has even better color?

As for face-specific processing: e.g. performing extra NR and/or lower contrast on face to reduce wrinkles and imperfections: it wouldn't be a bad option as long as it can be turned off.

In the video I posted with EOS-R, A7 III, XT-3 using basically auto everything, the Fuji looks the most accurate/realistic. Canon is classically too red/yellow (which is R+G) and Sony is way too yellow (R+G): neither is accurate. Fuji could use a hair less magenta, but it's decent as is (better than a green bias). That being said, those of us with the experience and inclination can tweak in-camera color and get much better results (as well as post treatment). What I really liked about the 5D Mark II and III was great color out of the camera (a bit too red but usable). Canon went backwards in color after the 5D3. Even with extensive tuning, I couldn't get the C300 II to match the tuned skin tones on the 1DX II. At this point I really don't want to spend time color tuning a new camera: it should be accurate using default settings (which is why ARRI is still #1 in the pro space).

Major kudos to Fuji with the XT-3- leap-frogging not only Panasonic and Sony, but also Canon for less money! Since Fuji makes a MF body, curious if they'll ever make a FF body: very useful for stills regarding ease of super shallow DOF (FF lenses still provide most affordable options; already have many FF lenses); video it's cool but not as important (don't shoot ultra shallow DOF video very often).

If you look at even the AF video that was just posted (comparing 4 cams), the skin on the Fuji looks like some smoothing has been applied (in camera). It's consistent. I have the impression I am looking at an image where compression has been used to get rid of artifacts, but only where the skin is concerned. I think Fuji is baking something in with this, albeit quite successfully. It's not just about the color, it's about the texture of the skin... Look at 2:30 on the X-T3 vs Canon R video above for instance.

 

[marvinhello]

If you look at even the AF video that was just posted (comparing 4 cams), the skin on the Fuji looks like some smoothing has been applied (in camera). It's consistent. I have the impression I am looking at an image where compression has been used to get rid of artifacts, but only where the skin is concerned. I think Fuji is baking something in with this, albeit quite successfully. It's not just about the color, it's about the texture of the skin... Look at 2:30 on the X-T3 vs Canon R video above for instance.

It was shot with default NR setting at 0, with -4 being off.

 

[yoclay]

It was shot with default NR setting at 0, with -4 being off.

I am speaking about something related to the way skin is treated, rather than the overall image.
If most of these cams can do face detection, they can certainly do skin smoothing without affecting other aspects of the image.
Since the vast majority of what people shoot are faces, this would not surprise me in the least.

 

[jcs]

I am speaking about something related to the way skin is treated, rather than the overall image.
If most of these cams can do face detection, they can certainly do skin smoothing without affecting other aspects of the image.
Since the vast majority of what people shoot are faces, this would not surprise me in the least.

Yeah I noticed that in Max's video. Studying the paused faces at 2:30, it looks like the Fuji isn't in focus. Also note that F4 is not equivalent to full frame re: DOF (see bokeh differences for the non-crop FF 4K cameras). Fuji should have used F2.8. Also color is off (white is blue-ish), though overall skin tones look more accurate on the Fuji (and Nikon; white shirt looks yellowish on Sony+Canon and blue-ish on Fuji+Nikon; WB issues). There's also something going on with the Fuji- lower contrast and/or something with the gamma curve making the face look under exposed*. When looking at the Slashcam demo, the Fuji skin tones look really, really good compared to the EOS-R. At 2:37, the Fuji has more skin detail vs. the EOS-R.

A bit more research shows many other folks making the same statement: Fuji has passed Canon for color / skin tones. For stills, there's unfortunately not equivalent lenses for crop sensors. E.g for full frame we have the stellar Canon 70-200 F2.8 II/III, Fuji provides the 50-140 F2.8: it needs to be F2.8/1.6 = F1.8 to match FF (F2 would be close enough). Using a focal reducer and Canon/FF lenses is an option, though haven't found anything providing decent AF-C performance for video. Might be good enough for stills- need to do more research.

Was looking forward to the A7S III (still am); hope the XT-3 helps Sony de-gimp the A7S III as much as possible before launch (supposed to be officially announced any time now), most importantly color science. That means great skin tones straight from camera using default settings (no funky color bias's as still present on the A7 III (agreed can produce decent color with tweaking)).


* EDIT: whole frame is underexposed- compare background.

 

[David_Manning]

I can only share my personal opinions, of course...but I like the video coming from the XT-3 better than the video out of my Sony FS5. Naturally, the FS5 is better as an overall package FOR ME (XLR, locking SDI, and especially AutoND...another game-changer), but the XT-3 is smaller, lighter, has better AF for video, and shoots 4K like a real 4K camera, not like an HD camera that can also shoot 4K (FS5, I’m looking at you here).

As far as the video AF, I haven’t personally used Canon’s DPAF, but I can say that the Fuji’s eye-detection AF works incredibly well, allowing me to use my native X-mount Fujinon glass to shoot wide open at F1.4 and keep the subject nailed, even during movement. It’s like having a magic focus-puller. Using the “standard” video AF, it holds and tracks as well as I need, and being able to pull focus via the touch-screen is as good as I’ve seen on my friend’s 1DXII.

So, yes...I’m now part of the club, thrilled to be IN the club, and very happy I made the decision. In order to make this camera a world-beater, enable internal 10-bit 4:2:2 4K, add IBIS, and an AutoND like the Sony FS5, and it’d never leave my hands for any other camera.

 

[jcs]

Cool list of (probably) every lens you can use with the XT-3: https://alikgriffin.com/a-complete-list-of-fujifilm-x-mount-lenses/

If not worried about AF/OIS/Iris there's a lot of decent manual lenses (e.g. fast Voigtlander M's with an adapter will look excellent). Also some incredibly cheap lens option (fast too; sure there will be 'issues', let's call them 'character' ).

 

[Grug]

Alrighty, first 'studio' build complete!

Weight's coming in at 9.7kg/21lbs with the Zeiss 28-80mm CZ.2 up front, which feels quite nice - compared to around 14kg/30lbs for the same gack on my Sony F3, or 20kg/44lbs(!) on my Alexa.

There's perhaps a couple of hundred grams to be saved by moving to carbon fibre rods everywhere, and a more lightweight connection to hold the Odyssey7Q on there.

With the single 150W/hr v-mount powering everything, we're looking at 6hrs(!) of continuous runtime. Which makes me feel all warm and fuzzy inside.

The camera is powered through the DC port on the grip, via a Wooden Camera Canon C100 DTAP cable (which includes a 9v transformer, making it compatible with the Fuji).

The battery plates I've sourced are the Wooden Camera Pro models, which feature 3 built-in DTAP ports. That's great, because it allows me to power everything through the plate, and the single DTAP on the battery itself, without the need for any splitters.

Certainly an extra 500g could easily be saved by moving to a smaller battery. But I have a case full of these Anton Bauers for the Arri, so I don't think I'll be adding more to the fleet anytime soon.

And a lighter mattebox than the Arri MMB-2 would shave some weight too. But I'm wary of using anything less solid, since it feels like having the rod-mounted MMB-2 supporting the front of the lens, is the only thing preventing it from ripping the lens mount out of the camera!

Feeding the HDMI signal to the Odyssey allows me to run SDI to the Teradek (which is working fine) and the Gratical EVF.

I'm regretting having sold off my Odyssey7Q+ though, because the 7Q doesn't support 4k HDMI. Which means either adding a 4k HDMI-to-SDI converter to all of the other gack (if I want 4k 4:2:2 via the 7Q), or buying another recorder :/ though for the time being, having 10-bit 1080p 4:2:2 external, and 10-bit 4k 4:2:0 internal makes me feel reasonably well covered for recording.

My thoroughly mediocre cable management gives me pause. But a good AC will get that sorted out nicely I reckon.

On the whole, I'm pretty happy with the build. I have some more parts on the way to refine it a little further. And I might add rods on the assistant's side of the camera (simply to make the camera a little easier to grab).

What do people think?