Radical feature improvement - subject tracking with manual focus (Nikon Z f state of the art)

[Guest]

Advertisement (gone after registration)

3 minutes ago, pgk said:

The look at the plethora of other cameras which do work better for you. The M Rangefinder is what it is and no amount of tinkering with it can effectively modify it into being something it is not. At best it would be a hobbled version of whatever electronic camera it would try to mimick. At worst such software intrusions would be an unweildy mess. Leave well alone and use it as designed.

I don't agree. There is a medium-effort engineering ask to Leica to make the M work much better with some of the most expensive M lenses that Leica is selling. Why would they not want to take advantage of this opportunity?

[jaapv]

1 minute ago, mzbe said:

'Select Subject'. Has been there for a couple of years ...

What has that to do with focusing?

[Anbaric]

With this system, of course, we shouldn't be satisfied with a single rangefinder patch. Instead, we should demand a matrix of 24 separate rangefinder patches arranged in a 6x4 grid so that we can use the one the computer identifies as covering the subject, ending the misery of central focusing and recomposing. It's what Oskar Barnack would have wanted.

[Guest]

2 minutes ago, charlesphoto99 said:

Leica 135mm with an M9, full frame. It's possible. 

 

So what? I have similar pictures taken with the rangefinder. There are circumstances where a percentage of images will be sharp.

Please note it's not just about focus ... Even if you use the RF for framing and focus, this could improve metering hugely - by giving you spot metering via subject detection and automatic selection of 'point of interest' (not going to call it AF point to avoid confusion). Metering is one of the not-so-great areas of the M11, no?

[Guest]

5 minutes ago, jaapv said:

What has that to do with focusing?

The ability for software (M11 firmware, if Leica would do this) to select an area of interest based on 'flat' image data.

My point is that there is zero impediment, aside from the willingness to do this and the effort required, for Leica to retrofit this feature with the M11.

[Guest]

Good discussion all around. I value your perspective. Clearly, we have different needs.

Since none of us control what Leica does, let's leave it here?

[jaapv]

Advertisement (gone after registration)

Just now, mzbe said:

I don't agree. There is a medium-effort engineering ask to Leica to make the M work much better with some of the most expensive M lenses that Leica is selling. Why would they not want to take advantage of this opportunity?

Medium effort? The only thing Nikon did was to shut off the pulses from the existing AF system to the lens motor and replace it by the fingers of the photographer. Leica would have to implement a complete AF system in the M to implement something like this

[charlesphoto99]

3 minutes ago, mzbe said:

So what? I have similar pictures taken with the rangefinder. There are circumstances where a percentage of images will be sharp.

Please note it's not just about focus ... Even if you use the RF for framing and focus, this could improve metering hugely - by giving you spot metering via subject detection and automatic selection of 'point of interest' (not going to call it AF point to avoid confusion). Metering is one of the not-so-great areas of the M11, no?

I would have missed that shot if I was having to pay attention to crap bouncing all over a screen in order to focus. I've used the new and old Visoflex (and LV on the back) and it makes me want to puke. Same with Q's, SL's, and other mirrorless brands. To me the M is about purity of working in analog. And that may mean some missing some things, which AF has done to me more often than not when I've used it (Nikon). I use my knowledge of light for spot metering (and with digital we have immediate confirmation vs waiting for the negs/slides to come back from the lab). 

I'll bet though that Leica are working on this exact kind of stuff for the future. It seems to be what people want. But to me, with more and more technology, we've become 'prisoners of our own convenience.' 

[jaapv]

19 minutes ago, mzbe said:

So what? I have similar pictures taken with the rangefinder. There are circumstances where a percentage of images will be sharp.

Please note it's not just about focus ... Even if you use the RF for framing and focus, this could improve metering hugely - by giving you spot metering via subject detection and automatic selection of 'point of interest' (not going to call it AF point to avoid confusion). Metering is one of the not-so-great areas of the M11, no?

With me, about 90% will be in focus, even on the M8 where it is an effective 200. What is wrong with metering on the M11? Exposure is determined by the photographer, not the camera.

This all comes from the mentality, fed by present-day automation, that the craft of photography should be shifted away from the photographer to the camera.

 

17 minutes ago, mzbe said:

The ability for software (M11 firmware, if Leica would do this) to select an area of interest based on 'flat' image data.

My point is that there is zero impediment, aside from the willingness to do this and the effort required, for Leica to retrofit this feature with the M11.

Err... I'll select the area of interest myself, if you don't mind.

[jaapv]

4 minutes ago, charlesphoto99 said:

I'll bet though that Leica are working on this exact kind of stuff for the future.

I suppose they have to to sell cameras and turn a profit.

[Guest]

7 minutes ago, jaapv said:

Medium effort? The only thing Nikon did was to shut off the pulses from the existing AF system to the lens motor and replace it by the fingers of the photographer. Leica would have to implement a complete AF system in the M to implement something like this

Courtesy of Chat GPT

• Explain how a camera uses subject detection in software. Please state if there are hardware dependencies on additional sensors or optics.

ChatGPT

ChatGPT

Subject detection in camera software is an advanced feature that leverages a combination of hardware and software capabilities to identify and focus on specific subjects within a frame, such as people, animals, or objects. Here's an overview of how it works and the potential hardware dependencies:

1. Image Sensor and Processor: The heart of subject detection lies in the camera's image sensor and processor. The sensor captures the light information, which is then processed by the processor. Modern sensors are capable of capturing high-resolution images, which provide detailed data for accurate subject detection.

2. Software Algorithms: The processor runs software algorithms that analyze the image data. These algorithms are designed to recognize patterns and shapes that correspond to common subjects. For instance, face detection algorithms look for oval shapes with two symmetrical dark areas (eyes) and a triangle (nose).

3. Machine Learning and AI: Many modern cameras incorporate machine learning and AI to improve subject detection. These systems are trained on vast datasets of images to recognize a wide variety of subjects under different conditions. AI can adapt to different lighting, angles, and even partial obstructions.

4. Hardware Dependencies:

   • Additional Sensors: Some cameras, especially those in smartphones, may use additional sensors like infrared, LiDAR, or depth sensors to assist in subject detection. These sensors provide additional data points about the distance and shape of objects in the frame, enhancing the accuracy of detection, especially in low light or complex scenes.
   • Optics: The quality of the camera's lens can impact subject detection. A higher quality lens can produce clearer, more detailed images, making it easier for the software to identify subjects. However, the lens itself does not actively participate in the subject detection process; it's more about the quality of the input data.

5. Autofocus Integration: Subject detection is often tied to the camera's autofocus system. Once a subject is detected, the camera can automatically adjust focus to ensure that the subject is sharp. This is particularly important in photography involving moving subjects.

6. Real-time Processing: Modern cameras can perform subject detection in real time, providing feedback to the photographer or automatically adjusting settings for optimal shots. This requires significant processing power and efficient software.

In summary, subject detection in cameras is primarily a software-driven process, but it greatly benefits from high-quality hardware, including advanced image sensors, additional depth or IR sensors, and quality optics. The integration of AI and machine learning has significantly enhanced the capabilities and accuracy of subject detection systems in modern cameras.

[jaapv]

And now chat GPT is a photographer and camera designer? Ending with a spot of advertising for AI? 🤣🤣🤣

[Guest]

8 minutes ago, charlesphoto99 said:

I would have missed that shot if I was having to pay attention to crap bouncing all over a screen in order to focus. I've used the new and old Visoflex (and LV on the back) and it makes me want to puke. Same with Q's, SL's, and other mirrorless brands. To me the M is about purity of working in analog. And that may mean some missing some things, which AF has done to me more often than not when I've used it (Nikon). I use my knowledge of light for spot metering (and with digital we have immediate confirmation vs waiting for the negs/slides to come back from the lab). 

I'll bet though that Leica are working on this exact kind of stuff for the future. It seems to be what people want. But to me, with more and more technology, we've become 'prisoners of our own convenience.' 

I would propose to agree on an 'AND' - you can have your cake, and I can eat mine? It's not a zero sum game ...

[charlesphoto99]

4 minutes ago, mzbe said:

I would propose to agree on an 'AND' - you can have your cake, and I can eat mine? It's not a zero sum game ...

Sorry if you feel like we're giving you a hard time. It's just utterly unrealistic what you're proposing in the current form the M exists in. I guess some of us old timers would prefer to see the M refined as an analog beast that happens to shoot digital vs shoehorning more and more tech into it that compromises things like startup time, reliability etc. 

[charlesphoto99]

And then the subject covered his eyes and the camera went into a complete meltdown.... 😏

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!

[jaapv]

10 minutes ago, Al Brown said:

Current AI's overuse of words to describe something is a big form of verbal smoke and serves as a means of masking the lack of understanding or knowledge on the topic. It stems from countless instances in real life, mostly corporate presentations (ant politics lol), where the speaker attempts to enforce a sense of authority and expertise through the (over)use of technical terms, without actually providing any valuable practical information. AI algorithms have definitely fallen prey to this pattern of behavior during their learning. The true challenge lies in the ability to "filter out the noise"...

The correct term is "gobbledegook"  🙃

[jaapv]

36 minutes ago, mzbe said:

Courtesy of Chat GPT

• Explain how a camera uses subject detection in software. Please state if there are hardware dependencies on additional sensors or optics.

 

ChatGPT

ChatGPT

Subject detection in camera software is an advanced feature that leverages a combination of hardware and software capabilities to identify and focus on specific subjects within a frame, such as people, animals, or objects. Here's an overview of how it works and the potential hardware dependencies:

1. Image Sensor and Processor: The heart of subject detection lies in the camera's image sensor and processor. The sensor captures the light information, which is then processed by the processor. Modern sensors are capable of capturing high-resolution images, which provide detailed data for accurate subject detection.

2. Software Algorithms: The processor runs software algorithms that analyze the image data. These algorithms are designed to recognize patterns and shapes that correspond to common subjects. For instance, face detection algorithms look for oval shapes with two symmetrical dark areas (eyes) and a triangle (nose).

3. Machine Learning and AI: Many modern cameras incorporate machine learning and AI to improve subject detection. These systems are trained on vast datasets of images to recognize a wide variety of subjects under different conditions. AI can adapt to different lighting, angles, and even partial obstructions.

4. Hardware Dependencies:

   • Additional Sensors: Some cameras, especially those in smartphones, may use additional sensors like infrared, LiDAR, or depth sensors to assist in subject detection. These sensors provide additional data points about the distance and shape of objects in the frame, enhancing the accuracy of detection, especially in low light or complex scenes.
   • Optics: The quality of the camera's lens can impact subject detection. A higher quality lens can produce clearer, more detailed images, making it easier for the software to identify subjects. However, the lens itself does not actively participate in the subject detection process; it's more about the quality of the input data.

5. Autofocus Integration: Subject detection is often tied to the camera's autofocus system. Once a subject is detected, the camera can automatically adjust focus to ensure that the subject is sharp. This is particularly important in photography involving moving subjects.

6. Real-time Processing: Modern cameras can perform subject detection in real time, providing feedback to the photographer or automatically adjusting settings for optimal shots. This requires significant processing power and efficient software.

In summary, subject detection in cameras is primarily a software-driven process, but it greatly benefits from high-quality hardware, including advanced image sensors, additional depth or IR sensors, and quality optics. The integration of AI and machine learning has significantly enhanced the capabilities and accuracy of subject detection systems in modern cameras.

 

The problem is more basic: The feature uses AF data and software to implement. The M11 has no AF data and software.

[charlesphoto99]

As it is, Leica rarely, if almost never, add new 'features' to existing cameras once the production model is set. They will clean up bugs, maybe adjust an existing setting to be less awkward, etc. But suddenly taking firmware that is of another platform entirely and adding it to the M is never going to happen. Not even Nikon etc does this - i.e. the Z50 is the same basic platform as the Z9. They're not taking bits of Z software and then adding it to the D850 or D500. Firmware is tightly tied to camera hardware. No amount of wishful thinking will create a unicorn. The small team of coders Leica has are most likely still debugging the M11 and working on the next slew of M,Q, and SL releases. There will always be new models with new features for those afflicted with GAS to trade in to.

[Warton]

I recently acquired a Nikon Zf and was amazed by how well its auto-focus works. The subject recognition tracks subject accurately and swiftly, with the inexpensive Nikon Z mount 40mm/f2 SE lens.

I don't understand German but this video makes me interested in a VC 40mm f/1.2.

Having said I have no interest of this kind of feature in my M bodies. I keep rangefinder as a....rangefinder and never use EVF anyway. If I want Zf type of mirrorless I grab my Nikon.

[Advertisement]

Hallo,

Look here: Interesting Leica gear

[pgk]

1 hour ago, mzbe said:

My point is that there is zero impediment .....

Many of us buy the M RF because of its simplicity. Adding features distracts from this. Its pretty straightforward.