SL2?

[frame-it]

Advertisement (gone after registration)

2 hours ago, Chaemono said:

One advantage of the algorithms taking over the ‘photography’ world is that they will likely conduct these forum discussions less emotionally. 😁

An AI algorithm might actually ignore all the humans ;)

[hillavoider]

I quite like this digital and film option 

https://www.hasselblad.com/cfv-ii-50c-907x/

 

 

[bags27]

4 hours ago, hillavoider said:

I quite like this digital and film option 

https://www.hasselblad.com/cfv-ii-50c-907x/

 

 

+1 they are doing very creative things. Of course, it's an old school tripod camera. And the digital back effectively replaces the old Polaroid to give an instant read before using film. Very cool!

Edited June 21, 2019 by bags27

[BernardC]

On 6/20/2019 at 7:54 AM, Chaemono said:

They are not well equipped for AI and Computational Imaging. Basically, algorithms will create beautiful landscape photography. 

Oh please, that's wrong both coming and going.

Leica's partnership with Huawei means that they are arguably better equipped for "computational photography" than any other photographic brand (including Apple).

That being said, we've been reading about how "computers will create perfect pictures," or music, or books, or food, for three quarters of a century now. The best we've ever gotten from that is pale lifeless imitations of the real thing.

Here's a typical example:

They applied cutting-edge techniques to calculate/create a new Rembrandt painting. That's a very limited data set compared to "landscape photography" (for instance), and they applied way more processing power at it than you will see inside a camera for a decade or more.

What they created is possibly the most mediocre Rembrandt ever. Something that one of his students might throw together if he was short on cash... In other words, if that was the best Rembrandt could do, nobody would remember who Rembrandt was.

That's what you get with huge resources and a very limited mandate. They copied a well-documented style that was popular centuries ago, and that was a failure. How will "computational photography" anticipate the next trends? Things change from month to month, it's no use making your pictures look like something from the past.

[Chaemono]

vor einer Stunde schrieb BernardC:

..., we've been reading about how "computers will create perfect pictures," or music, or books, or food, for three quarters of a century now.

Yes, concepts like AI aren’t new, but we’re getting closer to this becoming a reality. Thanks to GPGPU there’s been a logarithmic increase in compute power and a logarithmic drop in compute cost. See here: https://wwwmayr.in.tum.de/konferenzen/Jass09/courses/2/Puzyrev_paper.pdf. And this was 2009. 

To illustrate this graphically see here: 

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

The Human Genome Project was started in 1990 and was finished in 2003. It cost over $3b. 16 years later the whole thing (sequencing all base pairs that make up human DNA) only costs $1,000 and the total expense will drop to $100 at some point in the near future. 

Edited June 22, 2019 by Chaemono

[Tailwagger]

Building a neural network can indeed require huge computation cost to create the model, but the result of that calculation can typically execute on any run of the mill CPU.  Personally, and I tangentially work in this space, I think we're a long way from replacing ourselves, not that it isn't inevitable one way or another.

[Chaemono]

Advertisement (gone after registration)

vor 5 Stunden schrieb Tailwagger:

Personally, and I tangentially work in this space, I think we're a long way from replacing ourselves, not that it isn't inevitable one way or another.

For many tasks, yes. We’re a along way from replicating the intelligence found in the tiniest natural brains even. Bees, for example, forage up to four miles (6.374 km) from their hive and find their way back even though they don’t have a GPS module. 😁

[jaapv]

Well, they do have solar navigation by polarized light.

[gpwhite]

On 6/16/2019 at 11:19 AM, Chaemono said:

I had rented the X1D with the 45 and the 90. I found the XCD lenses a bit soft wide open, plus one gets those freakin octagonal bokeh balls. In the link are some test shots vs. the M10 with the Summicrons: https://www.smugmug.com/gallery/n-9FSKSS/i-Qxw7zjd. I didn't buy it and lost the rental fee. I found it slow and I could focus faster with the M10. And the Summicrons stopped down a bit were stunning.

I'll rent the X1D Mk 2 with a couple of lenses eventually and compare it to the SL2, whenever that comes out, with the SL Summicrons, so some of the finest lenses vs. some of the finest lenses wide open. 😁 Of course, stopped down the XCD lenses are excellent, as are most lenses.

Try out the XCD 65mm, and I expect you will be fully satisfied. Rendering is lovely, close focus superb, and open or down 1/2-stop the aperture reflection into your bokeh is just as round as you APO 50 Summicron.

My conclusion is that even the weaker XCD lenses (i.e., 45mm) are not soft, but the slow capture system in the X1D and demands of a 50Mp sensor bring no margin for error in your triggering technique.

[Flat Earth]

On 6/22/2019 at 9:05 AM, Chaemono said:

The Human Genome Project was started in 1990 and was finished in 2003. It cost over $3b. 16 years later the whole thing (sequencing all base pairs that make up human DNA) only costs $1,000 and the total expense will drop to $100 at some point in the near future. 

The bottleneck in genome sequencing is no longer the cost of the sequencing. This is why you see the sequencing cost per genome levelling off in 2015. For the moment, there is no economic benefit to making it cheaper. The issue is computational.  A human genome is four billion bases long but is currently sequenced in 200 base pair fragments. Error correction and the correct assembly of these pieces is the labour and cost-intensive part because this is not yet effectively automatable. $100 human genome will come but it's going to be slower than you think. 

[tom0511]

vor 6 Stunden schrieb gpwhite:

Try out the XCD 65mm, and I expect you will be fully satisfied. Rendering is lovely, close focus superb, and open or down 1/2-stop the aperture reflection into your bokeh is just as round as you APO 50 Summicron.

My conclusion is that even the weaker XCD lenses (i.e., 45mm) are not soft, but the slow capture system in the X1D and demands of a 50Mp sensor bring no margin for error in your triggering technique.

If you shoot the 90 wide open and have the right settings the bokeh is not hexagonal (after the firmware updates). I am also a long time M user and x1d user. I can see how the Leica is more compact, faster to startup and no blackout at all. I can not confirm the xcd lenses to be soft, indeed I believe they are sharper and the larger sensor also leads to a less abrupt falloff from focus plane to background. Some of the Leica M lenses show a smoother bokeh though. 

[ynp]

2 hours ago, tom0511 said:

If you shoot the 90 wide open and have the right settings the bokeh is not hexagonal (after the firmware updates). I am also a long time M user and x1d user. I can see how the Leica is more compact, faster to startup and no blackout at all. I can not confirm the xcd lenses to be soft, indeed I believe they are sharper and the larger sensor also leads to a less abrupt falloff from focus plane to background. Some of the Leica M lenses show a smoother bokeh though. 

Excuse my ignorance, I have no firsthand experience with the X1D. I have always thought that the bokeh is defined by the lens and the shape of the aperture. How it’s possible to fix the bokeh by a firmware fix? The firmware makes the hexagonal bokeh round in camera?

 

 

[ericborgstrom]

7 hours ago, ynp said:

" …The firmware makes the hexagonal bokeh round in camera?"

 

 

Yes, the apperture blades now opens up extra at maximum making it round. At smaller openings the shape is still hexagonal.

[Bob Andersson]

12 hours ago, ynp said:

Excuse my ignorance, I have no firsthand experience with the X1D. I have always thought that the bokeh is defined by the lens and the shape of the aperture. How it’s possible to fix the bokeh by a firmware fix? The firmware makes the hexagonal bokeh round in camera?

While waiting for the camera to ship I'm doing the sad thing! 🤣 From page 103 of the X1D II manual in the Camera Configuration Settings section:

Max Aperture Setting

Select if the lens shall use a fully open and round aperture at maximum setting. A round aperture will create a smoother look for the out-of-focus areas. For some lenses, e.g. the XCD 90, this can cause a very slight overexposure.

Normal

Standard setting. Will minimize the risk for internal reflections.

Full

Selects a fully open round aperture.
Only for XCD Lenses.

The manual also includes a couple of images illustrating the difference which I won't post here but the difference is significant. Downloadable from the My Hasselblad section of the Hasselblad website.

[Chaemono]

vor 8 Stunden schrieb Bob Andersson:

While waiting for the camera to ship I'm doing the sad thing! 🤣 From page 103 of the X1D II manual in the Camera Configuration Settings section:

Max Aperture Setting

Select if the lens shall use a fully open and round aperture at maximum setting. A round aperture will create a smoother look for the out-of-focus areas. For some lenses, e.g. the XCD 90, this can cause a very slight overexposure.

Normal

Standard setting. Will minimize the risk for internal reflections.

Full

Selects a fully open round aperture.
Only for XCD Lenses.

The manual also includes a couple of images illustrating the difference which I won't post here but the difference is significant. Downloadable from the My Hasselblad section of the Hasselblad website.

So the X1D has three wide open settings, Normal wide open, Full wide open, and Max Aperture Setting. 😂 

[FlashGordonPhotography]

3 hours ago, Chaemono said:

So the X1D has three wide open settings, Normal wide open, Full wide open, and Max Aperture Setting. 😂 

No. Just two. Max Aperture Setting is the menu title. The other two are the options.

Gordon

[gpwhite]

I would just go with the 65mm and call it a day.

The 90mm XCD is nice, but you do see the compression of the focal length.

Trying to follow up on Tom0511 very insightful comments,  I would agree that XCD are "sharper" than many Leica M lenses of similar focal length. But rendering is not as good, other than the 65mm. For example, I prefer the look of my Elmarit-M 24mm ASPH to my "sharper" XCD 30mm. But the deciding factor for me has been the exceptional color and plasticity of the XCD files. M files are very good, but the XCD files are just a cut above....  often, no changes to color or levels are needed on the RAW, but just a tweak of shadows or EV depending upon where you selected the exposure setting.

[meerec]

4 hours ago, gpwhite said:

would just go with the 65mm and call it a day.

I’d go with the 80/1.9 and call it a day 🙂 🤗

[hoppyman]

On 7/4/2019 at 8:29 AM, ynp said:

Excuse my ignorance, I have no firsthand experience with the X1D. I have always thought that the bokeh is defined by the lens and the shape of the aperture. How it’s possible to fix the bokeh by a firmware fix? The firmware makes the hexagonal bokeh round in camera?

 

 

I think they are describing the difference when the aperture blades are fully retracted, that is the aperture is effectively round, not limited by the shape formed by the edges of the aperture blades 

[Advertisement]

Hallo,

Look here: Leica L-Mount

[ynp]

46 minutes ago, hoppyman said:

I think they are describing the difference when the aperture blades are fully retracted, that is the aperture is effectively round, not limited by the shape formed by the edges of the aperture blades 

Thank you, understood now. So it affects only the wide open shooting, looked at the flickr pictures and not many pictures have bad looking out focus areas.