LEDE ON
It’s quarterly earnings report time across the tech world. Fujifilm was the first I’ve seen in the camera biz to specifically call out “increases in semiconductor memory and IT-related component prices” as reason for their negative forward forecast. But given that there are reports that there are US$1b (yes billion) in Apple Silicon processors sitting at TMSC waiting for RAM to show up and be integrated, you have to think that every camera maker is in the same boat. Instead of waiting for Godot, we’re waiting for DRAM.
——————————
News
Sony Continues Releasing New Things
Seems that Sony felt mid-summer was the time to go on a product release spree. One thing about the slowdown in photography product introductions is that we don’t have companies stepping on each other’s press releases at the moment.
This week gave us the 100-400mm f/5.6-8 OSS lens, a US$850 consumer companion that slots well under the US$4300 100-400mm f/4.5GM that was also recently introduced. No G, slower aperture, and no tripod collar are the most noticeable differences, but 23 ounces (654g) is also quite light for a lens of this focal range. Morever, it’s very Tamron-like in its use of 67mm front filters. Surprisingly, you get a focus limiter switch, an OSS on/off switch, an Fn button, the lens works at up to 120 fps on the Sony bodies, and it also works with Sony’s 1.4x and 2x teleconverters (giving you 140-560mm f/8-11 and 200-800mm f/11-16, respectively).
Note that the aperture fall-off begins as early as 135mm, where the lens is already f/6.3. The f/8 aperture is reached at 220mm, well before you get to the end of the lens’ zoom range. On the other hand, the lens focuses reasonably close (34” or 0.86m at 400mm), which provides it near macro level of magnification (1:2.4). Inside are two ED and two aspherical elements, which are protected by dust and moisture resistant design.
So now the 100-400mm choices look like this:
- Canon 100-400mm f/5.6-8 IS (US$750)
- Canon 100-400mm f/4.5-5.6L IS II (US$2700)
- Leica 100-400mm f/5-6.3 (US$2600)
- Nikon 100-400mm f/4.5-5.6 VR S (US$2950)
- Panasonic 100-400mm f/4-6.3 OIS II (US$1700)
- Sigma 100-400mm f/5-6.3 OS (US$1050)
- Sony 100-400mm f/5.6-8 OSS (US$850)
- Sony 100-400mm f/4.5GM OSS (US$4300)
Technically there’s a Fujifilm 100-400mm f/4.5-5.6 OIS at US$2250, but since that’s specific for crop sensor use, it doesn’t exactly line up with the above list. Note also that Tamron has a 50-400mm f/4.5-6.3 at US$1300, which is my current preference for a more budget model with a 100-400mm focal range. Nikon is noticeable absent with a budget version, and the S version they do have is one of their earliest Z-mount lenses and could use a bit of a refresh given all the recent competitor offerings.
——————————
News
Adobe Chugs Along
The just-released Lightroom Classic 15.5 got the ACR feature/edge sliders for AI masks and a few other changes including Generative Expand for Lightroom (not Classic). Denoise AI finally made it to the iPad Lightroom. ACR itself got a new blemish remover. Adobe also introduced a new plugin for ChatGPT, allowing that AI to manage and use the Creative Cloud’s products from within the AI itself.
That latter bit, by the way, is the most likely way you’ll be using AI for photography in the future. Products such as Photoshop have gotten so complex that typing “drop the bottom stop of dynamic range in this image more quickly to black” in a chat engine sure beats trying to remember which Photoshop tools you should be using, finding them, and manipulating them directly.
While I don’t use AI to create anything new (writing, photos, spreadsheets, etc.), I do use it after the creation to do some button up operations that would take me more time to do by hand. Indeed, over time you start building up .MD (markdown) files that can be used to sequence an AI through the things you want done (e.g. add a watermark, add Copyright information, output as a PNG, submit the PNG file to Squash to reduce the file size for the Web, add the bythom- prefix to the name, and put the result in the correct Web image resource folder). Think of this type of AI use more as automation of steps and sequences than doing something on its own. It really does make you quite productive once you master it.
If you’re wondering how this is all done, the big AI companies now support something called MCP (Model Context Protocol). I’m amused, as this mimics what we were doing with objects in the PenPoint Operating System back in the early 90’s (everything you programmed in Penpoint was understood by everything else others were doing, allowing new features and processes to be added to everything simultaneously, and all data available everywhere). Basically, MCP is a communication system that has been standardized. So Excel, Photoshop, et.al., now expose externally everything they can do, and the AI engines then can use that to control applications and automate processes. MCP also extends into data storage and online tools, as well, so every day we see more and more things that can managed through the MCP.
——————————
Commentary
No Log Jam Here
We’re officially well past the half-way point of the year. For photography enthusiasts, it has been slim pickings for new cameras so far this year. To wit:
- Canon R6 V
- Leica SL3-P
- Panasonic L10
- Ricoh GRIV Monochrome
- Sony a7R VI
- Sony FX5
- Sony RX10 V
Technically, only two of those sit in the main mirrorless branch (SL3-P and a7R VI). The remainder are two “creators” cameras with emphasis on video, and three compact cameras, though one of them is not so compact. Nothing from Fujifilm, Nikon, or OMDS.
At the top end of mirrorless we now have this situation:
- Canon R1 is two years old.
- Fujifilm X-H2S is four years old.
- Nikon Z9 is nearly five years old.
- OMDS OM-1 II is over two years old.
- Sony A1 II is nearly two years old.
At the bottom end of mirrorless things look worse:
- Canon RS100 is over three years old.
- Fujifilm X-T30 III is just less than a year old.
- Nikon Z5030 is four years old.
- OMDS OM-D E-10 IV is six years old.
- Sony A6100 is seven years old.
In Japanese Consumer Electronics fashion, the top end of cameras define what follows in terms of technology and features for the entire lineup. Nikon is starting to become a top-end lagger (after being a leader), but at the lower end of lineup, both OMDS and Sony are lagging the laggard.
The real question here is whether or not there’s really anything worth iterating any more, though.
First, let me repeat what I’ve always written about digital (dating back well more than 30 years): I’ll always take more sampling, all else being equal (or at least nearly so). The two places that shows up the most is in dynamic range and pixel count (some say frame rate and other things, too). For dynamic range, the sampling extends the capture range, while for pixels it adds additional sub-sampling most of the time, which helps with aliasing (few are using added pixel counts to make larger prints). I’ll take both advances, as long as nothing else suffers.
But I’m in the Ansel category, where I pursue every small bit of benefit. Optimal capture, processed optimally (hey, I got it down to four words this time; usually it takes me six ;~).
I’m doubting that even a small portion of the customer base for cameras would take the same stance as me. A really solid 20” print takes just 24mp, and 95%+ of scenes can have their dynamic range accurately rendered if you’re exposing correctly. On the video side, 4K is ubiquitous and all you need (particularly if you’re capturing and outputting in HDR), as there's very few benefits from capturing more pixels.
So the connundrum for camera makers is…actually, it’s multiple simultaneous problems now:
- Supply chain constraints force camera makers into less iteration. Memory is sold out now through the end of 2027, so if you need more for a particular new camera or just need more to sell more cameras, not going to happen.
- Economics force camera makers into less iteration. Memory prices increased by 2x+, other semiconductor prices jumped far higher, as well. Margins on lower end products are squeezed very tight now, and payback of R&D costs takes more time.
- Rights issues keep top-end still cameras from transitioning to video cameras to get higher frame rates. Do pro sports photographers need more than 20 fps? No. It just increases their workflow, and they can’t use the 30 fps+ output as video due to video rights being owned by others.
- Long-term customers already have highly competent products. The five-year old Z9 is still an extremely competitive camera, and does pretty much everything a Nikon user might want. Sure, a Z9II will have a lot of updates, but the only one that would trigger a lot of updates is much better autofocusing.
- Smartphones, with care, can do big prints and feature film output. Here we are almost 20 years post iPhone, and the camera makers still haven’t figured how to market against smartphones. Or integrate with them. Meanwhile, the volume of smartphones sold so far eclipses dedicated cameras, new imaging tech now shows up in phones first.
So where exactly is the “opportunity” to grow?
The Chinese have shown both Japan (compact, ILC) and the US (GoPro-type action cameras) that some of that opportunity centers around the Apple Think Different notion. Witness the camera-on-a-gimbal-stick variants that have taken off (e.g. DJI Osmo Pocket).
One issue is that the camera market was already very niche—Canikony divide it up into 10 or more very similar models each—so if you’re not first into a new niche that appears (or is forced), then you have a difficult time getting any traction, as Nikon discovered with the KeyMission cameras.
I’d judge that pretty much any new dedicated camera model needs a minimum of 100k lifetime likely unit sales to even get off the drawing board. Anything less than that becomes too much R&D and marketing expense to get a decent ROI (return on investment). These days, the camera makers need more ROI than before, because they can no longer get near-zero percent loans in Japan. It doesn’t help that the Japanese camera makers are great at the R&D but not so good at the marketing, either. Tokyo almost requires that a product go viral on launch now, as the on-going marketing will be minimal and ineffective. That sets a high bar for any new camera product.
A number of Nikon users wonder where the Z7III is. The problem there is that the Z8 is effectively what a Z9-generation Z7III would be if they just iterated “as normal” (and now priced at that price point). Sony started having the same problem with the A7R series, which is why there’s now a different body design in their lineup, the A7CR. Sony has more unit volume than Nikon, which allows them to sub-niche themselves some. A Z7III as most users would expect it to be, on the other hand, would make Z8 sales collapse.
Small niches do exist that could be targeted. Just as I outlined in the first decade of this century, dedicated black and white and infrared are two such sub-niches. That seems to be working to extend Ricoh’s small-scale efforts (and Leica’s), but probably wouldn't work for Canikony. Again, they really need 100k+ expected units to wake up an engineering team from their slumber.
So what’s happening now is an introduction slowdown. At one point in the digital camera era, compacts and low-end consumer ILC iterated on 12-month boundaries, prosumer at 24, and top-end pro at 48, and they did so with predictable regularity. While the top-end likely won’t stretch too much longer—though Nikon will likely hit 60 months with their top-end Z9 iteration—it’s the lower models that are seeing the big gaps now, with 36 to 48 months being more the norm now than the exception.
No worries, mate. Just go out and photography with what you have.