One last X-Trans Vs. Bayer comparison

[Erik Baumgartner]

Here are 2 images to compare, both are tweaked for max clean detail as I would normally do for any image (not default values), the XH-1 (X-Trans) using Lightroom and Iridient X-Transformer, and The X-T100 (Bayer) with just Lightroom. If you carefully look around the images you will see that the Bayer sensor is bit better in some areas - fine text especially, and the X-Trans sensor in others - check the green stuff and the peacock feather. To my eye there is no clear overall winner here. As I wanted to compare finished results, I did clean up some moire on both files - they both had a little, but in different places, the X-Trans was only a tiny bit better here. Easy fix in both cases. I don't see any differences in color here that couldn't easily be adjusted to match the other (both were processed using the same tweaked version of the Adobe Color profile). Is the more complicated X-trans processing really worth it anymore?

X-H1 ISO 200 - X-Trans with Lightroom/Iridient X-Transformer

X-T100 ISO 200 - Bayer with Lightroom

 

[tokumeino]

Do you mean "yet another" comparison ? Because I'm afraid that it won't be the last !

 

[Erik Baumgartner]

Do you mean "yet another" comparison ? Because I'm afraid that it won't be the last !

Agreed, 'yet another' would have probably have been a better choice.

 

[Advent1sam]

Here are 2 images to compare, both are tweaked for max clean detail as I would normally do for any image (not default values), the XH-1 (X-Trans) using Lightroom and Iridient X-Transformer, and The X-T100 (Bayer) with just Lightroom. If you carefully look around the images you will see that the Bayer sensor is bit better in some areas - fine text especially, and the X-Trans sensor in others - check the green stuff and the peacock feather. To my eye there is no clear overall winner here. As I wanted to compare finished results, I did clean up some moire on both files - they both had a little, but in different places, the X-Trans was only a tiny bit better here. Easy fix in both cases. I don't see any differences in color here that couldn't easily be adjusted to match the other (both were processed using the same tweaked version of the Adobe Color profile). Is the more complicated X-trans processing really worth it anymore?

Thanks Erik, Agreed, but we are stuck with xtrans for the higher end bodies currently and I actually think the x-h1 is the best of the breed with the caveats above, to me the 24mp xtrans looks better than the x-t3! Can you share your settings for both these images, incl xtransofrmer for the x-h1, what's your modified color profile?

X-H1 ISO 200 - X-Trans with Lightroom/Iridient X-Transformer

X-T100 ISO 200 - Bayer with Lightroom

 

[rwbaron]

Interesting and honestly, I enjoy these so thanks.

When I played with ID I found it did extract more fine X-Trans detail but at the expense of adding moire' which I see in your conversion using IXT. I can read slightly further into the black, grey and white text boxes with the Bayer version but IMO that's the only area I see an advantage for Bayer. Everywhere else I look it's either a wash or a slight advantage for X-Trans. Surprisingly the X-Trans greens are better (look at the foliage and especially the lower contrast green backing board). Also look at the pencil sketch of the group sitting looking at the portrait. There are pencil lines that are not visible with the Bayer but clearly there with the X-Trans.

It all depends on where one looks and what converter is used for X-Trans. They both have their strengths and weakness and I don't see any real downside for X-Trans other than you have to use a different workflow to extract the absolute best. That said, I find Lr perfectly satisfactory for 99% of what I shoot with X-Trans so for me it's not a big deal and all this is much ado about nothing really for the way most of these images are displayed. People make this out like their shots are all for display in a gallery at 40x60 inches.

Though I've not played with it yet it appears the latest Silky Pix release is impressive and could actually be a better option than IXT if one doesn't mind doing the conversion to a Tiff for further adjustments elsewhere.

Bob

 

[The Fat Fish]

Please bring out a Bayer X-T3!

 

[Erik Baumgartner]

Interesting and honestly, I enjoy these so thanks.

When I played with ID I found it did extract more fine X-Trans detail but at the expense of adding moire' which I see in your conversion using IXT. I can read slightly further into the black, grey and white text boxes with the Bayer version but IMO that's the only area I see an advantage for Bayer. Everywhere else I look it's either a wash or a slight advantage for X-Trans. Surprisingly the X-Trans greens are better (look at the foliage and especially the lower contrast green backing board). Also look at the pencil sketch of the group sitting looking at the portrait. There are pencil lines that are not visible with the Bayer but clearly there with the X-Trans.

It all depends on where one looks and what converter is used for X-Trans. They both have their strengths and weakness and I don't see any real downside for X-Trans other than you have to use a different workflow to extract the absolute best. That said, I find Lr perfectly satisfactory for 99% of what I shoot with X-Trans so for me it's not a big deal and all this is much ado about nothing really for the way most of these images are displayed. People make this out like their shots are all for display in a gallery at 40x60 inches.

Though I've not played with it yet it appears the latest Silky Pix release is impressive and could actually be a better option than IXT if one doesn't mind doing the conversion to a Tiff for further adjustments elsewhere.

Bob

Using IXT only for its demosaicing and with it's own sharpening and NR turned off, I see no difference in moire or anything else other than improved rendering of fine detail and the reduction of unsightly artifacts. Used selectively as a plug-in within LR, I find it effortless to insert into my workflow. Note, IXT helps, but doesn't actually make a big difference on this particular image.

While I agree that once you've sorted out how to get the get the most out of x-trans that there is no real downside to it, and I do slightly prefer the results I got out of the X-H1 image above, I don't see any major upsides to x-trans either. The X-T100 image is very, very close in terms of detail, at least as good in terms of color, and better with fine text and other similar fine structure. Remember, this is an entry level model we're talking about here, you could buy an X-T100 with your choice of some very premium lenses for the price of just an X-H1 body. If you try to reproduce the X-T100 image above using an x-trans file with only Lightroom, I think you will probably come up a bit short. I suspect that many photographers would be well served by a premium Fuji body with a Bayer sensor.

Here's the X-H1 RAF if anyone wants to have a go matching or bettering the X-T100 image above with just Lightroom (or whatever): https://www.dpreview.com/reviews/im...ge?s3Key=463b1fa3548f4ad99b6c0e26f56284d6.raf

 

[sankos]

I think you've put the X-T100 at a disadvantage here by using Adobe demosaicing. When I converted both X-T20 and X-T100 files in IXT and brought them both to Lr for sharpening, the latter was visibly more detailed.

Adobe demosaicing for Bayer cameras is hardly state-of-the-art. It's not particularly sharp, yet it has problems with moire. What helps them is the pretty good Sharpening module (esp. the smart Detail slider). Typically AMaZE does a much better job for most types of detail -- it's not perfect but I prefer it in many cases.

 

[Ben Herrmann]

...I'd have to say that in this case, the X-H1 wins in the detail department, as I see more details in the various scenes.

 

[Fezhat]

One last X-Trans Vs. Bayer comparison

baying for worms

 

[The Davinator]

XTrans does push above its weight.

 

[rwbaron]

Interesting and honestly, I enjoy these so thanks.

When I played with ID I found it did extract more fine X-Trans detail but at the expense of adding moire' which I see in your conversion using IXT. I can read slightly further into the black, grey and white text boxes with the Bayer version but IMO that's the only area I see an advantage for Bayer. Everywhere else I look it's either a wash or a slight advantage for X-Trans. Surprisingly the X-Trans greens are better (look at the foliage and especially the lower contrast green backing board). Also look at the pencil sketch of the group sitting looking at the portrait. There are pencil lines that are not visible with the Bayer but clearly there with the X-Trans.

It all depends on where one looks and what converter is used for X-Trans. They both have their strengths and weakness and I don't see any real downside for X-Trans other than you have to use a different workflow to extract the absolute best. That said, I find Lr perfectly satisfactory for 99% of what I shoot with X-Trans so for me it's not a big deal and all this is much ado about nothing really for the way most of these images are displayed. People make this out like their shots are all for display in a gallery at 40x60 inches.

Though I've not played with it yet it appears the latest Silky Pix release is impressive and could actually be a better option than IXT if one doesn't mind doing the conversion to a Tiff for further adjustments elsewhere.

Bob

Using IXT only for its demosaicing and with it's own sharpening and NR turned off, I see no difference in moire or anything else other than improved rendering of fine detail and the reduction of unsightly artifacts. Used selectively as a plug-in within LR, I find it effortless to insert into my workflow. Note, IXT helps, but doesn't actually make a big difference on this particular image.

Check the resolution targets at the 4 corners of the frame. IXT has introduced moire' on some of the elements that's not present with DPR's ACR version. I'm not saying it's an issue for all intents and purposes but just that it's there.

While I agree that once you've sorted out how to get the get the most out of x-trans that there is no real downside to it, and I do slightly prefer the results I got out of the X-H1 image above, I don't see any major upsides to x-trans either. The X-T100 image is very, very close in terms of detail, at least as good in terms of color, and better with fine text and other similar fine structure. Remember, this is an entry level model we're talking about here, you could buy an X-T100 with your choice of some very premium lenses for the price of just an X-H1 body. If you try to reproduce the X-T100 image above using an x-trans file with only Lightroom, I think you will probably come up a bit short. I suspect that many photographers would be well served by a premium Fuji body with a Bayer sensor.

Cant' argue with that but I think Fuji sees X-Trans as a differentiator that makes their system unique. Marketing is important. ;-)

 

[Erik Baumgartner]

Interesting and honestly, I enjoy these so thanks.

When I played with ID I found it did extract more fine X-Trans detail but at the expense of adding moire' which I see in your conversion using IXT. I can read slightly further into the black, grey and white text boxes with the Bayer version but IMO that's the only area I see an advantage for Bayer. Everywhere else I look it's either a wash or a slight advantage for X-Trans. Surprisingly the X-Trans greens are better (look at the foliage and especially the lower contrast green backing board). Also look at the pencil sketch of the group sitting looking at the portrait. There are pencil lines that are not visible with the Bayer but clearly there with the X-Trans.

It all depends on where one looks and what converter is used for X-Trans. They both have their strengths and weakness and I don't see any real downside for X-Trans other than you have to use a different workflow to extract the absolute best. That said, I find Lr perfectly satisfactory for 99% of what I shoot with X-Trans so for me it's not a big deal and all this is much ado about nothing really for the way most of these images are displayed. People make this out like their shots are all for display in a gallery at 40x60 inches.

Though I've not played with it yet it appears the latest Silky Pix release is impressive and could actually be a better option than IXT if one doesn't mind doing the conversion to a Tiff for further adjustments elsewhere.

Bob

Using IXT only for its demosaicing and with it's own sharpening and NR turned off, I see no difference in moire or anything else other than improved rendering of fine detail and the reduction of unsightly artifacts. Used selectively as a plug-in within LR, I find it effortless to insert into my workflow. Note, IXT helps, but doesn't actually make a big difference on this particular image.

Check the resolution targets at the 4 corners of the frame. IXT has introduced moire' on some of the elements that's not present with DPR's ACR version. I'm not saying it's an issue for all intents and purposes but just that it's there.

Well it's microscopic, but I stand corrected, there is indeed a touch of moire in the corners that isn't present in the Adobe version. There is also, I might add, a better rendering of fine detail that might have something to do with that.

While I agree that once you've sorted out how to get the get the most out of x-trans that there is no real downside to it, and I do slightly prefer the results I got out of the X-H1 image above, I don't see any major upsides to x-trans either. The X-T100 image is very, very close in terms of detail, at least as good in terms of color, and better with fine text and other similar fine structure. Remember, this is an entry level model we're talking about here, you could buy an X-T100 with your choice of some very premium lenses for the price of just an X-H1 body. If you try to reproduce the X-T100 image above using an x-trans file with only Lightroom, I think you will probably come up a bit short. I suspect that many photographers would be well served by a premium Fuji body with a Bayer sensor.

Cant' argue with that but I think Fuji sees X-Trans as a differentiator that makes their system unique. Marketing is important. ;-)

I again agree with you on this, but I'm thinking that maybe it's use as a market differentiator has run its course. It is entirely likely that pool of the camera buyers who are willing to deal with the extra hassle of X-trans processing, be it real or perceived, is dwindling. I've seen lots of potential new customers recently who appear to be jumping off on the other side of the Fuji/everybody else fence for just that reason. Maybe it's time for Fuji to start thinking beyond x-trans. Perhaps they're holding out until some new unique sensor tech becomes available. I guess we'll have to wait and see.

 

[Tom Schum]

I developed the raws using the free Fuji/Silkypix raw developer v3.0, all at default settings. This software is available here:

RAW FILE CONVERTER EX 3.0 powered by SILKYPIX

Download the latest version of RAW FILE CONVERTER EX 3.0 powered by SILKYPIX for FUJIFILM X Series and GFX.

www.fujifilm.com

X-H1


X-T3


X-T100

--
Tom Schum
Copper: Mankind's favorite electrical conductor

 

[57even]

Interesting and honestly, I enjoy these so thanks.

When I played with ID I found it did extract more fine X-Trans detail but at the expense of adding moire' which I see in your conversion using IXT. I can read slightly further into the black, grey and white text boxes with the Bayer version but IMO that's the only area I see an advantage for Bayer. Everywhere else I look it's either a wash or a slight advantage for X-Trans. Surprisingly the X-Trans greens are better (look at the foliage and especially the lower contrast green backing board). Also look at the pencil sketch of the group sitting looking at the portrait. There are pencil lines that are not visible with the Bayer but clearly there with the X-Trans.

It all depends on where one looks and what converter is used for X-Trans. They both have their strengths and weakness and I don't see any real downside for X-Trans other than you have to use a different workflow to extract the absolute best. That said, I find Lr perfectly satisfactory for 99% of what I shoot with X-Trans so for me it's not a big deal and all this is much ado about nothing really for the way most of these images are displayed. People make this out like their shots are all for display in a gallery at 40x60 inches.

What, you mean they're not? ;-)

Though I've not played with it yet it appears the latest Silky Pix release is impressive and could actually be a better option than IXT if one doesn't mind doing the conversion to a Tiff for further adjustments elsewhere.

It is indeed very good. Best colour of all the raw converters IMO, and a very natural look.

Big difference from v2, and the updated HDR function is excellent.

 

[57even]

I think you've put the X-T100 at a disadvantage here by using Adobe demosaicing. When I converted both X-T20 and X-T100 files in IXT and brought them both to Lr for sharpening, the latter was visibly more detailed.

Adobe demosaicing for Bayer cameras is hardly state-of-the-art. It's not particularly sharp, yet it has problems with moire. What helps them is the pretty good Sharpening module (esp. the smart Detail slider). Typically AMaZE does a much better job for most types of detail -- it's not perfect but I prefer it in many cases.

In which case, there is no disadvantage to Xtrans at all.

The workarounds I use for Xtrans are the same as those I used for my D800.

 

[Erik Baumgartner]

I developed the raws using the free Fuji/Silkypix raw developer v3.0, all at default settings. This software is available here:

http://www.fujifilm.com/support/digital_cameras/software/myfinepix_studio/rfc/


X-H1


X-T3


X-T100

You would think that Fuji's own converter would closely match the SOOC JPEG, but it's quite a bit off (below for comparison)


X-H1 ISO 200 SOOC JPEG

 

[Hitch22]

1/80 sec v 1/60 sec...there is a brightness difference.

 

[stevo23]

Interesting and honestly, I enjoy these so thanks.

When I played with ID I found it did extract more fine X-Trans detail but at the expense of adding moire' which I see in your conversion using IXT. I can read slightly further into the black, grey and white text boxes with the Bayer version but IMO that's the only area I see an advantage for Bayer. Everywhere else I look it's either a wash or a slight advantage for X-Trans. Surprisingly the X-Trans greens are better (look at the foliage and especially the lower contrast green backing board). Also look at the pencil sketch of the group sitting looking at the portrait. There are pencil lines that are not visible with the Bayer but clearly there with the X-Trans.

It all depends on where one looks and what converter is used for X-Trans. They both have their strengths and weakness and I don't see any real downside for X-Trans other than you have to use a different workflow to extract the absolute best. That said, I find Lr perfectly satisfactory for 99% of what I shoot with X-Trans so for me it's not a big deal and all this is much ado about nothing really for the way most of these images are displayed. People make this out like their shots are all for display in a gallery at 40x60 inches.

Though I've not played with it yet it appears the latest Silky Pix release is impressive and could actually be a better option than IXT if one doesn't mind doing the conversion to a Tiff for further adjustments elsewhere.

Bob

Using IXT only for its demosaicing and with it's own sharpening and NR turned off, I see no difference in moire or anything else other than improved rendering of fine detail and the reduction of unsightly artifacts. Used selectively as a plug-in within LR, I find it effortless to insert into my workflow. Note, IXT helps, but doesn't actually make a big difference on this particular image.

Check the resolution targets at the 4 corners of the frame. IXT has introduced moire' on some of the elements that's not present with DPR's ACR version. I'm not saying it's an issue for all intents and purposes but just that it's there.

It is - good catch. However, it's at the next to smallest spatial frequency on that chart and I don't see it in the woodcut. So not bad at all. If it takes that to see that tiny bit, we're doing well!

While I agree that once you've sorted out how to get the get the most out of x-trans that there is no real downside to it, and I do slightly prefer the results I got out of the X-H1 image above, I don't see any major upsides to x-trans either. The X-T100 image is very, very close in terms of detail, at least as good in terms of color, and better with fine text and other similar fine structure. Remember, this is an entry level model we're talking about here, you could buy an X-T100 with your choice of some very premium lenses for the price of just an X-H1 body. If you try to reproduce the X-T100 image above using an x-trans file with only Lightroom, I think you will probably come up a bit short. I suspect that many photographers would be well served by a premium Fuji body with a Bayer sensor.

Cant' argue with that but I think Fuji sees X-Trans as a differentiator that makes their system unique. Marketing is important. ;-)

 

[Badwater]

From the samples on the YOuTube, Fuji always improves their sensors. Their newer sensors look as if they outperform the older ones. Take for example the new highly popular and highly rated X-T3. From what I see, it's the best APSC Crop on the market for video and photo. No other crop camera can match it's performance in video or photo: from what I see on the YouTube, and Flickr. Whatever the sensor is, the processor has the power to make it the best. And that's pretty much all there is to it.


I've looked at countless 4K video, and pics, and what people do with the X-T3 is amazing. The results of what skilled shooters can do with this camera makes it the best APSC IMO.