RoyC
Active member
Pretty much the same here I have a 180-400 and 800 that are going nowhere.
Official Nikon Z9 Launch, Info, and Discussion Thread
- Thread starter Steve
- Start date
If you would like to post, you'll need to register. Note that if you have a BCG store account, you'll need a new, separate account here (we keep the two sites separate for security purposes).
Pretty much the same here I have a 180-400 and 800 that are going nowhere.
I can only hope that Nikon introduces Z replacement for the 180-400 and 600. The latter with a drop in TC.
The 600 is on the roadmap so probably will hear something about it 2022. But I totally agree with you that a built in TC would be a killer feature.
thelordofthelight
Well-known member
600 is already in the roadmap so it's just a matter of few months but I have my doubts on lenses like 180-400/120-300 etc. The Z9 seems to have addressed one of the biggest concerns/ debated topics regarding Z Mount, which is the performance of F mount glass. I never thought the Z9 would all of a sudden make F mount glass work so seemlessly. That's why I feel Nikon may not prioritise launching similar focal lengths that are already in f mount ( except the most used ones like 2.8 Trinity, 400/500/600 etc.) The PF options in the new roadmap also seems to send a message that Nikon does not want over lapping lenses in both the mounts (500pf in f mount vs 400/800 in Z mount).
John Navitsky
Well-known member
I haven’t watched this video yet, but when I test drove the a1 with dog sports i observed a variety of shifts to body, or toy, etc. as well. Dogs are a hard subject.
Hope the 600 w/ TC comes in 2022, but I suspect not. Product announcement for the 400 w TC had a development announcement. Time in development? 3, 6, 9 months?
I would rather not use FTZs. Two step to connect a lens. Plus storing / carrying a lens in a camera bag fits better w/o a FTZ. I can barely fit my 600 in my camera bag, add a FTZ it will not fit. the 180-400 (200-400) fits attached to camera body (dSLR) will fit. Not sure about the size (length) of a Z9 (or Z6, Z7) w/ 180-400 vs Dxxx w 180-400.
O
Well-known member
That's more anxiety excitement LOL, i go to sleep in a few minutes, i sleep lightly, but well, i usually wake up at 2 or 3 minutes before the alarm.
Its not uncommon for some people, if your relaxed and do meditation in different ways it helps.
The mind can behave like a delinquent child or puppy that loves being addicted to behavioral patters............so its important to adopt the right ones.
I am a heavy sleep apnia person ans use a CPAP machine with great success, my AHI reading is 1 now which is excellent, used to be 54, your lungs clean the blood with oxygen while you sleep, this helps reduce toxins and supports the organs greatly in doing their job, the end result is you get god sleep, your body is tuned while you sleep, when you add meditation to the equation you end up in control of your mind rather than it controlling you LOL
All fun......happy days.
sid_19911991
Well-known member
These 2 blokes attended an on-site Z9 presentation gathering & they were told by the presenter that the Z9 AF system (3D tracking in particular) does not use contrast detect system but tracks subjects based on colour profiles/colour patterns.
Any idea how A1 & R5/R3 goes about it?
Any idea how A1 & R5/R3 goes about it?
sid_19911991
Well-known member
The 1-stop lower dynamic range performance is not a problem for me as it is still close to the best (A7 iii 14.7 ETC. ) & much better than Canon R6 (13.4) A9 (13.3). which is terrible IMO. (Source
XO Mark)What concerns me is the ISO performance on the Z9.
Canon R6 is wrongly mentioned on the main page I guess.
sid_19911991
Well-known member
Nice review.
Loved the video footage of the geese/swans flying.
Some takeaways from the basic review.
Rolling shutter on the 4k 120 is lower relatively.
AF for birding can improve further with u
Video af is perhaps the best in the market now!
ISO performance not great. My biggest concern.
Nice review.
Loved the video footage of the geese/swans flying.
Some takeaways from the basic review.
Rolling shutter on the 4k 120 is lower relatively.
Stills AF for birding can improve further with updates.
Video af is perhaps the best in the market now!
ISO performance not great. My biggest concern.
O
Well-known member
The 1-stop lower dynamic range performance is not a problem for me as it is still close to the best (A7 iii 14.7 ETC. ) & much better than Canon R6 (13.4) A9 (13.3). which is terrible IMO. (Source
What concerns me is the ISO performance on the Z9.
Canon R6 is wrongly mentioned on the main page I guess.
View attachment 28711
Hence i said a few days ago that IMO its logical there will be a super high ISO lower res King released by someone if not Nikon once the Z9 flow has filled the pipeline.
Its inevitable by manufactures choice that one camera will not do it all. Sony has the A1 and A9 A 7 Nikon has the D850 D6............its a fair assumption.........now there is also some Nikon propaganda that HI ISO is possibly a thing of low priority going forward and they will not need to focus on that in the future..........could it be that 8k video and 120 fps is the foundation to start building the path of migration to get stills from video, it seems the way the industry is creeping to.
At what point will the Video component out weigh still photography as a component in a camera...........i mean IMO Video has seen a bigger growth in the past few years than stills.
The new technology emerging IMO will also see 80mp to 100mp landscape version arise at some point in time..........i smell Canon is working on that ? Sony is at 61mp already ?
Now don't get me wrong i am not saying there is anything wrong with the Z9, The Z9 is excellent, its basically a D850 Z7II on steroids with a new focusing system, great speed, new sensor, good video......and that fills a huge THIRSTY market.
Is the Z9 No 1 only from what i have seen and read its close to Sony and Canon which is awesome and i cant wait to get my Z9.
Nikon products are always under rated at launch............they do have teh best image files.
There are new options/technology emerging all the time, the Z9 is real a welcome one.........for now...........
What do i think of the eye tracking face tracking focus recognition and all that regardless of brand, be it Sony Canon or Nikon, from what i read see and hear, excellent, if not some what over cooked by marketing. Good one Madeleine Kay.....LOL
Only and opinion..........
sid_19911991
Well-known member
Yeah it remains to be seen what direction companies take.
Canon has a dedicated low light performer in R3 though. It is Canon's 1st BSI sensor & is rated the best camera for low light performance by DXO mark. A7 iii held that title previously, & having used the A7 iii, I can trust their opinion on the R3.
The high resolution SOny cameras (A7 r iii & A7 R iv) seem to have better ISO performance than the Z7 ii, while the Canon has a dedicated low light performer in R3. Z6 ii performs well in low light but its AF is not great & it has that stupid AA filter. The A7 iii & iv & R6/R5 etc does not have one, I have mo idea why Nikon bothered with a heavy AA filter on the Z6. Hope Nikon comes up with a 33mp low light beast with Z9 AF.
O
Well-known member
There is a gap in the market of chess, someone will have to fill it...........the Z9 is brilliant just an extension of the D850 Z7II in the next generation....
Contrast detect has been repalced by colour detect in the Z9 wonder how it works in mono ? can you even shoot in mono ? or even DX mode i haven't looked.
thelordofthelight
Well-known member
Z9 uses Hybrid phase-detection/Contrast AF. I think the 3D mode also uses additional details like colors, patterns etc. in addition to hybrid phase detect.
thelordofthelight
Well-known member
Having or not having an AA filter depends on the target audience for a particular camera. A lot of us (wildlife, macro, landscape) prefer not having an AA filter which is why Nikon has cameras like the Z7/9, D850 etc. There are folks who find moiré a deal breaker and for them not having an AA filter maybe a deal breaker.
sid_19911991
Well-known member
Yes, but apparently moire on the latest sensors are not a big deal at all! A lot of pros use non aa filter cameras like A7 iii, Z7 etc in fashion, weddings etc where moire could be an issue. That's what makes it puzzling.
fcotterill
Well-known member
Independent testing of 3D Tracking by Thom Hogan of the D5 revealed this mode uses colour information on chosen target. D6 probably works in similar ways but it also uses eye-tracking.
It's unclear if the Z9 is making use of colour patterns for its 3D tracking but this is not unlikely. We know enough so far that the shape recognition modes use a nested system of deep-learning algorithms to classify subjects hierarchically. This seems to be almost entirely on geometry: body > head/helmet > eyes/headlights.
The eye tracking works on marble statues and birds with uniform plumage eg crows, and so basically plumage of similar tones to the eyes. Examples of this ability also include mallard drakes. This implies the eye-recognition is using mostly if not entirely trained pattern recognition algorithms.
It's unclear if the Z9 is making use of colour patterns for its 3D tracking but this is not unlikely. We know enough so far that the shape recognition modes use a nested system of deep-learning algorithms to classify subjects hierarchically. This seems to be almost entirely on geometry: body > head/helmet > eyes/headlights.
The eye tracking works on marble statues and birds with uniform plumage eg crows, and so basically plumage of similar tones to the eyes. Examples of this ability also include mallard drakes. This implies the eye-recognition is using mostly if not entirely trained pattern recognition algorithms.
sid_19911991
Well-known member
The 3D tracking on the Z9 does not use contrast detect but uses colour pattern AI recognition as per the two blokes (refer to the link in the recent pages) who attended a Nikon Z9 presentation in a store. Well that's they were told they claim
There seems to have been some confusion related to this contrast statement. My guess is the following:
Remember that there are two tasks here: selecting what to focus on, and focusing on it. The "does not use contrast but colour" seems to be related to focus point selection, where it doesn't use contrast in the selection algorithm, but instead patterns. This is why in the first teaser (if people remember it), it was focusing on the model's eyes even with the extra white dots.
Once focus point is selected, there's the task of actual focusing. Given that the hybrid PD and contrast detect algorithm was working well, it could be that this is left unchanged, with maybe there relative weight of the two bring changed (and is pure PD in AF-C, as before). Pattern and colour for focusing is not useful, as far as I know.
My 2c...
Remember that there are two tasks here: selecting what to focus on, and focusing on it. The "does not use contrast but colour" seems to be related to focus point selection, where it doesn't use contrast in the selection algorithm, but instead patterns. This is why in the first teaser (if people remember it), it was focusing on the model's eyes even with the extra white dots.
Once focus point is selected, there's the task of actual focusing. Given that the hybrid PD and contrast detect algorithm was working well, it could be that this is left unchanged, with maybe there relative weight of the two bring changed (and is pure PD in AF-C, as before). Pattern and colour for focusing is not useful, as far as I know.
My 2c...
sid_19911991
Well-known member
"Remember that there are two tasks here: selecting what to focus on, and focusing on it."
In 3D tracking, the shooter manually decides what to focus on doesn't he/she? Remember I am talking about 3 D tracking alone & not 3D tracking with subject detection switched on.
Then once any random thing is selected, you are saying that the AF uses colour profiles to identify it...
Well, if we are manually selecting a random subject why does it need a colour profile recognition to identify it if it is not going to use colour profiling to track it.
Then you opine that the actual tracking from here on depends on phase detect & contrast detect & that using colour to track is not something you have never heard off before...
I have no clue about that. It will be nice to hear about such things from a Nikon technician.
fcotterill
Well-known member
Exactly, 2 tasks.... the Z9 is running multiple analyses in parallel on a moving subject: shape/pattern recognition to identify the head or helmet, or eyes or headlights etc, and at the same time it is keeping sharp focus on this selected target on the subject.
AI and DeepLearning are distinctly different. AI refers to an open system which improves its skillset based on new experiences, including new patterns it encounters that have taught its neural networks. DeepLearning - aka machinelearning - is the research/tech that encodes a decision-making algorithm into a device (eg Camera firmware) which in the the case of the Z9 comprises its finite repertoire of shape recognition modes. It is clear Nikon engineers structured this AF mode to use a nested system of deep-learning algorithms to classify subjects hierarchically.
Similarly to the DSLRs, D5 and D6 etc do, the Z9 may well use colour for tracking, as these 2 blokes report (above). This makes senses to detect a soccer player against background and players with different colours. However, the demonstrated examples of the Z9 with 3D tracking turned on, is this mode is tracking objects of many different colours: hence my post above arguing that pattern recognition seems to be fundamental.
AI and DeepLearning are distinctly different. AI refers to an open system which improves its skillset based on new experiences, including new patterns it encounters that have taught its neural networks. DeepLearning - aka machinelearning - is the research/tech that encodes a decision-making algorithm into a device (eg Camera firmware) which in the the case of the Z9 comprises its finite repertoire of shape recognition modes. It is clear Nikon engineers structured this AF mode to use a nested system of deep-learning algorithms to classify subjects hierarchically.
Similarly to the DSLRs, D5 and D6 etc do, the Z9 may well use colour for tracking, as these 2 blokes report (above). This makes senses to detect a soccer player against background and players with different colours. However, the demonstrated examples of the Z9 with 3D tracking turned on, is this mode is tracking objects of many different colours: hence my post above arguing that pattern recognition seems to be fundamental.
sid_19911991
Well-known member
I don't have a computer science background, I thought Deep Learning is the umbrella term for AI, Machine Learning, etc.
Actually it's the other way around with AI being the generic term that serves as the umbrella where deep learning is a subset of what would be considered AI. Here is a pretty good article in terms that do not require a CS degree. I do have a Computer Science degree but it is from the dark ages when AI was a concept and something talked about in computer labs. (ie. late 1980's)
Clearing the Confusion: AI vs Machine Learning vs Deep Learning Differences
Raise your hand if you’ve been caught in the confusion of differentiating artificial intelligence (AI) vs machine learning (ML) vs deep learning (DL)… Although the three terminologies are usually…
towardsdatascience.com
fcotterill
Well-known member
thanks, I posted links here on previous page. Although Nikon remains coy about the innards of deep-learning algorithms, Canon rhapsodized about how their's works: well, at least about some of the ways....
EDIT this interesting paper published a few months ago explains how a solution for camera AF can use DeepLearning (a Machine-Learning ML system that has been taught from reference databases). This solution exploits contrast detection signals; such that "deep learning control can achieve the scene-based advantages of contrast optimization at the real-time single frame rates of phase detection. Deep learning further enables dynamic control to image 3D and dynamic scenes in full focus."
ieeexplore.ieee.org
This is very unlikely to be Nikon's own solution but they admit to using trained neural networks to implement the Deep Learning of the subject recognition modes in a dedicated processor (namely the EXPEED7).
We know Nikon has invested in at least one DL/AI partner - wrnch - for detecting human poses and motion etc, which is presumably applicable to adapt to other organisms/3D objects. Previously, they bought out Mark Roberts Motion Control for their cutting edge tech in robotics
EDIT this interesting paper published a few months ago explains how a solution for camera AF can use DeepLearning (a Machine-Learning ML system that has been taught from reference databases). This solution exploits contrast detection signals; such that "deep learning control can achieve the scene-based advantages of contrast optimization at the real-time single frame rates of phase detection. Deep learning further enables dynamic control to image 3D and dynamic scenes in full focus."
Deep Learning for Camera Autofocus
Most digital cameras use specialized autofocus sensors, such as phase detection, lidar or ultrasound, to directly measure focus state. However, such sensors increase cost and complexity without directly optimizing final image quality. This paper proposes a new pipeline for image-based autofocus...
ieeexplore.ieee.org
This is very unlikely to be Nikon's own solution but they admit to using trained neural networks to implement the Deep Learning of the subject recognition modes in a dedicated processor (namely the EXPEED7).
We know Nikon has invested in at least one DL/AI partner - wrnch - for detecting human poses and motion etc, which is presumably applicable to adapt to other organisms/3D objects. Previously, they bought out Mark Roberts Motion Control for their cutting edge tech in robotics
Actually it's the other way around with AI being the generic term that serves as the umbrella where deep learning is a subset of what would be considered AI. Here is a pretty good article in terms that do not require a CS degree. I do have a Computer Science degree but it is from the dark ages when AI was a concept and something talked about in computer labs. (ie. late 1980's)
Clearing the Confusion: AI vs Machine Learning vs Deep Learning Differences
Raise your hand if you’ve been caught in the confusion of differentiating artificial intelligence (AI) vs machine learning (ML) vs deep learning (DL)… Although the three terminologies are usually…
Last edited:
I'm not hearing that at all. Brad Hill's testing of ISO and noise was pretty extensive and he considered the D850, Z7ii, and Z9 to be equal with minimal differences. He has done a comparison with a pre-production camera and firmware 1.0. It was in-depth comparing the Z6ii, Z7ii, D6 and D850 with the Z9 at all ISO levels.