Longer Lens vs. Teleconverter for Nikon Z

[adamhiram]

he didn't spend 5k on a lens, he spent more like $500 on a teleconvertor

Correct. While I would love a 120-300 f/2.8 or 400 f/2.8, $10k+ to photograph my kid playing soccer isn't in the budget.

 

[greybeard41 14]

@greybeard Think you might have misunderstood, he didn't spend 5k on a lens, he spent more like $500 on a teleconvertor.

As a general rule upscaling quality doesn't equal the IQ of a longer focal length. The advent of AI has improved the process, but still IMO there are advantages to a longer focal length.

imho, and experience, I have gotten better results by cropping and upscaling than by using a Nikon 1.4 teleconvertor which adds 1 stop more noise or 1 stop slower shutter speed than simple cropping. This was with a 70-200 f/2.8 and a Nikon D850. As to this "general rule" you speak of, I must not have gotten that memo.

 

[smoke665]

imho, and experience, I have gotten better results by cropping and upscaling than by using a Nikon 1.4 teleconvertor which adds 1 stop more noise or 1 stop slower shutter speed than simple cropping. This was with a 70-200 f/2.8 and a Nikon D850. As to this "general rule" you speak of, I must not have gotten that memo.

Upscaling uses algorithms to increase an image's resolution, by inventing new pixels based on surrounding pixels. AI uses artificial intelligence and deep learning to add better guess what a pixels should be. The downside is that it can introduce visual distortions such as blurriness, smearing, or "ghosting," particularly with fast-moving objects or very low base resolutions and can never be as sharp as detail captured at its native resolution.

As to noise, increasing the focal length, (even with a teleconvertor) doesn't increase noise, lowering the SNR does. Understanding where noise comes from and exposing for a full data file is the solution. See a simple explanation What Is Noise in Photography?

 

[greybeard41 14]

Upscaling uses algorithms to increase an image's resolution, by inventing new pixels based on surrounding pixels. AI uses artificial intelligence and deep learning to add better guess what a pixels should be. The downside is that it can introduce visual distortions such as blurriness, smearing, or "ghosting," particularly with fast-moving objects or very low base resolutions and can never be as sharp as detail captured at its native resolution.

As to noise, increasing the focal length, (even with a teleconvertor) doesn't increase noise, lowering the SNR does. Understanding where noise comes from and exposing for a full data file is the solution. See a simple explanation What Is Noise in Photography?

What I write about is from my experience, not what I have read online.

 

[smoke665]

What I write about is from my experience, not what I have read online.

And you assume mine isn't? I started this journey in the 60s. Experience without verification or factual basis is only an opinion.

 

[greybeard41 14]

And you assume mine isn't? I started this journey in the 60s. Experience without verification or factual basis is only an opinion.

Exactly, I've used a TC1.4 with a afs 70-200 f/2.8 and a afs 200-500. I prefer the cropped results over the TC, especially when upscaled with Topaz. Like any other tool, upscaling can be overdone leading to poor results. This has been my experience.

 

[smoke665]

, I've used a TC1.4 with a afs 70-200 f/2.8 and a afs 200-500. I prefer the cropped results over the TC, especially when upscaled with Topaz. Like any other tool, upscaling can be overdone leading to poor results. This has been my experience.

Like any tool TC's have advantages/disadvantages based on other factors. Putting a cheap TC between a good lens and sensor degrades the IQ. Putting a good TC between a not so great lens and sensor exaggerates the lens flaws. Mating a quality TC with a quality lens will generally increase the detail captured by the sensor, requiring less cleanup post. The caveat to that is a fully exposed newer high resolution sensor captures so much detail that even after cropping away pixels the remaining detail could be very close to the TC, the question then becomes one of remaining resolution. Adam is shooting a Z6II with a sensor resolution of 6048 x 4024 pixels, cropping away 28% of that for an equivalent FOV of 280mm focal length would leave 4233 x 2897, good enough for a 300 dpi 8x10, but not an 11x14. Upscaling it to the original resolution would require the algorithm to generate 28% of the pixels. The new AI features are good, but the less they have to guess at the better. To print a 16x20 with the 1.4 TC he'd need to upscale the original by 24% but to print a 16x20 with a crop to the equivalent FOV he'd need to upscale by 40% using your method. Care to guess how good that would look?

Choosing the right tool means considering all the factors required for the shot, experience comes into play when analyzing those factors. In the case of @adamhiram he tends to be very methodical in his write ups and comparisons, I would find it very hard to dispute his findings. Perhaps, if you'd like to start a new thread using the same approach, it might be interesting to all.