p.4 #2 · June 2026 LRC/PS updates ... faster Denoise measurements for Mac M with Neural Engine
davidrwilliams wrote:
[
Previous run of same with prior Lightroom version without ANE was about 22 seconds per image.
Strange .... Your Z9 photos (47Mpix) are completed in 9.96 sec / image, the A74 (61Mpix) is completed in 21 sec.?
30% more pixels, more than 100% extra time ?
Are you sure about both numbers? Then we have something strange at hand .... :-o
p.4 #7 · June 2026 LRC/PS updates ... faster Denoise measurements for Mac M with Neural Engine
I've been using this now for maybe a week, and the speed improvement is actually pretty amazing.
Before the update I was assuming that sharpening a 40MP raw file was going to take roughly 30 seconds, though sometimes it took a bit longer — the app was occasionally predicting 40 seconds. (It is possible that the latter were on MP files from a different camera.)
When I ran AI Denoise on these files it has indicating "10 seconds," but it seemed faster than that, and in fact it was when I counted the seconds.
I just sharpened another of these files and the app indicated 7 seconds... and it was that.
7 seconds!
This is on a m4 Mini Pro with 48GB memory, 12 cores.2 TB SSD.
I know that hardly seems like a very powerful system to some of you, but it actually works very effectively, supporting 2 27" monitors, a whole bunch of external storage, etc.
p.4 #8 · June 2026 LRC/PS updates ... faster Denoise measurements for Mac M with Neural Engine
gdanmitchell wrote:
I've been using this now for maybe a week, and the speed improvement is actually pretty amazing.
Before the update I was assuming that sharpening a 40MP raw file was going to take roughly 30 seconds, though sometimes it took a bit longer — the app was occasionally predicting 40 seconds.
When I ran AI Denoise on these files it has indicating "10 seconds," but it seemed faster than that, and in fact it was when I counted the seconds.
I just sharpened another of these files and the app indicated 7 seconds... and it was that.
7 seconds!
This is on a m4 Mini Pro with 48GB memory, 12 cores,.2 TB SSD.
I know that hardly seems like a very powerful system to some of you, but it actually works very effectively, supporting 2 27" monitors, a whole bunch of external storage, etc. ...Show more →
...not all - it's a tiny powerhouse! It's amazing how much processing power is now available in such a small form factor.
p.4 #9 · June 2026 LRC/PS updates ... faster Denoise measurements for Mac M with Neural Engine
davidrwilliams wrote:
...not all - it's a tiny powerhouse! Just amazing how much processing power is now in such a small form factor.
I have been a Mac user since… 1985. At one point I had a relationship with Apple that gave me access to a lot of snazzy stuff from the company.
I remember several phases in the development of computers for endeavors like photography. (And audio, which is what I taught for a ong time.)
Initially the computers simply were not powerful enough to do serious photography. It just plain was not an option. then we started to get apps like Photoshop (and eventually better printers) and it became possible… but you had no illusions about doing this on lower end computer models. Back then it meant buying a fairly high end tower system, adding a ton of external storage (or, in some cases, internal storage), radically upgrading ram, and paying what were then exorbitant prices of large tube monitors.
At some point I realized that the better iMacs were sufficiently powerful for serious still photography work. The last one I had was a top-end 27” model with an upgraded processor and a lot of ram… and it worked well. I realized that we had gotten to the point that a well-spec’ed iMac worked really, really well for photography.
When that last iMac became too long in the tooth (it was an Intel processor model) I took a close look at the specs for the mini and was pretty sure that it would work well, and that turned out to be right.
While the lowest of the low end machines still doesn’t have enough memory, storage, or (arguably) speed for serious photography work, amazingly we are now at a point where a low end model like the Mini, if properly equipped, has plenty of power.
p.4 #10 · June 2026 LRC/PS updates ... faster Denoise measurements for Mac M with Neural Engine
To my thinking the mini’s are basically the MacBook in a different form. Personally I went with an M4 studio mainly for the added connectivity and configuration capabilitiy, plus a MacBook pro for the field — at the time I bought the Studio, the mini only had 2 TB ports and why I passed on it in favor of the Studio. That changed in the next product cycle and so I’d agree it’s a perfectly viable desktop editing machine. I seem to have a knack for entering any given market one iteration early 🤷🏼
p.4 #11 · June 2026 LRC/PS updates ... faster Denoise measurements for Mac M with Neural Engine
Jack Flesher wrote:
To my thinking the mini’s are basically the MacBook in a different form. Personally I went with an M4 studio mainly for the added connectivity and configuration capabilitiy, plus a MacBook pro for the field — at the time I bought the Studio, the mini only had 2 TB ports and why I passed on it in favor of the Studio. That changed in the next product cycle and so I’d agree it’s a perfectly viable desktop editing machine. I seem to have a knack for entering any given market one iteration early 🤷🏼
Hey, I arguably made the opposite mistake when I got my last iMac, which I think was something like a 2018 or 2019 17” model with all the snazzy, faster stuff.. the last Intel processor version that made. As Apple switched to the M processors, it fell behind a bit too quickly!
The current mini has excellent ports, especially if you connect it to the Apple 27” monitor (I have two), which provide a whole bunch of additional ports.
p.4 #12 · June 2026 LRC/PS updates ... faster Denoise measurements for Mac M with Neural Engine
Jack Flesher wrote:
To my thinking the mini’s are basically the MacBook in a different form. Personally I went with an M4 studio mainly for the added connectivity and configuration capabilitiy, plus a MacBook pro for the field — at the time I bought the Studio, the mini only had 2 TB ports and why I passed on it in favor of the Studio. That changed in the next product cycle and so I’d agree it’s a perfectly viable desktop editing machine. I seem to have a knack for entering any given market one iteration early 🤷🏼
It used to be that a tower was a more powerful computer than a laptop. At least now with Apple's current 'M' chips, it's basically the same chip in different 'clothes'. You buy the clothes that suit your needs. Where there's maybe a slight benefit to a Mini or a Studio over a MBP or MBA, is thermal management and throttling, though apparently throttling is less problematic with these chips.
I probably would have bought a Mini if it came with a Max chipset. But they don't. I didn't want another laptop, already had a monitor, keyboard, etc., so got the Studio instead. And I wanted the Max for the 2x more GPUs over Pro options.
As we're seeing from the chart the OP has compiled, the newer chip generations bring some speed gains, even in the base or Pro configurations, that rival older Max systems.
p.4 #13 · June 2026 LRC/PS updates ... faster Denoise measurements for Mac M with Neural Engine
rscheffler wrote:
It used to be that a tower was a more powerful computer than a laptop. At least now with Apple's current 'M' chips, it's basically the same chip in different 'clothes'. You buy the clothes that suit your needs. Where there's maybe a slight benefit to a Mini or a Studio over a MBP or MBA, is thermal management and throttling, though apparently throttling is less problematic with these chips.
I probably would have bought a Mini if it came with a Max chipset. But they don't. I didn't want another laptop, already had a monitor, keyboard, etc., so got the Studio instead. And I wanted the Max for the 2x more GPUs over Pro options.
As we're seeing from the chart the OP has compiled, the newer chip generations bring some speed gains, even in the base or Pro configurations, that rival older Max systems....Show more →
I am not disappointed I did the Studio, in fact I would do it again. It’s fast, silent and runs cool essentially all the time, and increased processor and GPU definitely show their worth when processing with the newer algorithms. My only point was I could easily live with a mini if I were on a budget. Actually, knowing what I now know if I were on a budget, I would buy a decked out MBP and a dock for my monitor and external drive array and be happy with that. In fact, that may be where I start at the next upgrade. I’m 71 and am not sure how much longer I’ll be needing to keep current. At my age especially, good enough is usually plenty fast
p.4 #14 · June 2026 LRC/PS updates ... faster Denoise measurements for Mac M with Neural Engine
Jack Flesher wrote:
I am not disappointed I did the Studio, in fact I would do it again. It’s fast, silent and runs cool essentially all the time, and increased processor and GPU definitely show their worth when processing with the newer algorithms. My only point was I could easily live with a mini if I were on a budget. Actually, knowing what I now know if I were on a budget, I would buy a decked out MBP and a dock for my monitor and external drive array and be happy with that. In fact, that may be where I start at the next upgrade. I’m 71 and am not sure how much longer I’ll be needing to keep current. At my age especially, good enough is usually plenty fast ...Show more →
I could have afforded the Studio, but when I looked at the matter practically I realized that it didn’t provide any meaningful performance advantages. Again, I’m not exactly an entry level computer user… ;-)
Plus it is really great to have a computer that is so ridiculously tiny on my desktop. (I already had some experiences — both positive and negative — with earlier versions of the mini line that I had used as file servers.)
p.4 #15 · June 2026 LRC/PS updates ... faster Denoise measurements for Mac M with Neural Engine
gdanmitchell wrote:
I could have afforded the Studio, but when I looked at the matter practically I realized that it didn’t provide any meaningful performance advantages. Again, I’m not exactly an entry level computer user… ;-)
Plus is is really great to have a computer that is so ridiculously tiny on my desktop. (I already had some experiences — both positive and negative — with earlier versions of the mini line that I had used as file servers.)
I get it. Being retired and photo editing now just a hobby, the added speed is essentially irrelevant. But to be fair, in reality the Studio is only a few inches wider and deeper, and maybe an inch taller than the mini box. And again, really quiet. I don’t think there’s any wrong answer. Like I said, my next upgrade is likely a top end MBP and a docking station for my current desk setup, and done…
p.4 #16 · June 2026 LRC/PS updates ... faster Denoise measurements for Mac M with Neural Engine
Jack Flesher wrote:
I am not disappointed I did the Studio, in fact I would do it again. It’s fast, silent and runs cool essentially all the time, and increased processor and GPU definitely show their worth when processing with the newer algorithms. My only point was I could easily live with a mini if I were on a budget. Actually, knowing what I now know if I were on a budget, I would buy a decked out MBP and a dock for my monitor and external drive array and be happy with that. In fact, that may be where I start at the next upgrade. I’m 71 and am not sure how much longer I’ll be needing to keep current. At my age especially, good enough is usually plenty fast ...Show more →
I agree and that is basically what I did in 2022. Pandemic lockdowns pretty much wiped out my work for 2.5 years here in Canada but it resumed with a boom, I transitioned to mirrorless and my circa 2010 27" iMac i7 just wasn't cutting it any longer because of software gaps supporting new cameras I acquired. On a budget and practically starting over, I opted for a modestly performative MBP M1 Pro. I could dock it at home to an editing monitor and external drives while also enjoying the same processing power on the road. With the further addition of AI features in LR and especially my adoption of Denoise for practically everything, the MBP became less performative with camera updates that included 45MP files, up from 24MP. Last year I kind of lucked out finding a refurbished Studio M4 Max right about when I started shopping around for something more powerful than the MBP, for less than what it would have cost for the closest possible new Mini M4 Pro configuration.
During a recent roadtrip where I accompanied my wife for her work visiting various franchises she manages, I actually brought along the Studio and my editing display and either set it up at her site visits or in the hotel to plow through the ton of work I had on my plate. First time I ever did that with a 'tower' system. Mini would have been slightly easier to travel with, but as you mentioned, the Studio is only a inch or so larger on all dimensions and it nicely fit in the satchel I normally use to carry my MBP. It was the monitor that was the biggest hassle because I only had the box it originally came in (thankfully I hadn't discarded it) rather than a more space efficient carrying solution.
p.4 #17 · June 2026 LRC/PS updates ... faster Denoise measurements for Mac M with Neural Engine
Apple announced a new Mac Mini today and the base M6 now has dual 16-core Neural Engines. It will be interesting to see how this does with LR Denoise. It may get close to the M5 Pro/Max as LR uses the Neural Engine for Denoise.
p.4 #18 · June 2026 LRC/PS updates ... faster Denoise measurements for Mac M with Neural Engine
hoodlum90 wrote:
Apple announced a new Mac Mini today and the base M6 now has dual 16-core Neural Engines. It will be interesting to see how this does with LR Denoise. It may get close to the M5 Pro/Max as LR uses the Neural Engine for Denoise.
I will be very interested to see (though a MM is not in my future).
The MM is now available in 3 flavors;
M6 with 12 CPU / 12 GPU / Dual (2x?)16 NE
M5 Pro with 15 CPU / 16 GPU / 16 NE
M5 Pro with 18 CPU / 20 GPU / 16 NE
The correlation I ran on page 3 suggests that the GPU cores are the key with some influence based on technology node. On the new MM the M5 Pro chips both have an advantage for GPU core count but the M6 is a pretty significant design shrink factor from 3nm to 2nm. 32 NE cores have been available in the past on a Studio with an Ultra chip (essentially 2 Max chips attached together) and they always win, but also have 2x the GPU. So with 32 NE cores, but on a base M6 chip and only 12 GPU cores it will be interesting to see where this device falls for LRC/PS Denoise and other AI tasks.
https://www.macrumors.com/2026/08/25/apple-announces-2026-mac-mini/ "The M6 chip is equipped with a 12-core CPU and a 12-core GPU, up from a 10-core CPU and 10-core GPU in the M5 chip. In addition, the M6 chip features the first-ever dual Neural Engine with two 16-core engines, up from one in previous chips.
Apple said the Mac mini with the M6 chip delivers up to 40% faster CPU performance, up to 4x faster performance for AI tasks in particular, up to 2x faster graphics performance, and up to 2x faster storage speeds compared to the previous-generation model with the 10-core M4 chip, 32GB of unified memory, and 2TB of storage."
p.4 #19 · June 2026 LRC/PS updates ... faster Denoise measurements for Mac M with Neural Engine
If you need more oomph than the Mac Mini the Mac Studio was also updated today.
M5 Max 18 CPU / 32 GPU / 16 NE
M5 Max 18 CPU / 40 GPU / 16 NE
M5 Ultra 30 CPU / 64 GPU / 32 NE
M5 Ultra 36 CPU / 80 GPU / 32 NE
They can both be configured with up to 256GB RAM on current pre-orders but the Ultra will take 512GB in late October
The top-top-top config (which would include a 16TB SSD) will set you back a heartwarning $18,299 and that is before 512gb is available At $25/GB I think it will just topple over $25k including AppleCare if you find 256GB just won't do it ...
p.4 #20 · June 2026 LRC/PS updates ... faster Denoise measurements for Mac M with Neural Engine
mcbroomf wrote:
I will be very interested to see (though a MM is not in my future).
The MM is now available in 3 flavors;
M6 with 12 CPU / 12 GPU / Dual (2x?)16 NE
M5 Pro with 15 CPU / 16 GPU / 16 NE
M5 Pro with 18 CPU / 20 GPU / 16 NE
The correlation I ran on page 3 suggests that the GPU cores are the key with some influence based on technology node. On the new MM the M5 Pro chips both have an advantage for GPU core count but the M6 is a pretty significant design shrink factor from 3nm to 2nm. 32 NE cores have been available in the past on a Studio with an Ultra chip (essentially 2 Max chips attached together) and they always win, but also have 2x the GPU. So with 32 NE cores, but on a base M6 chip and only 12 GPU cores it will be interesting to see where this device falls for LRC/PS Denoise and other AI tasks.
https://www.macrumors.com/2026/08/25/apple-announces-2026-mac-mini/ "The M6 chip is equipped with a 12-core CPU and a 12-core GPU, up from a 10-core CPU and 10-core GPU in the M5 chip. In addition, the M6 chip features the first-ever dual Neural Engine with two 16-core engines, up from one in previous chips.
Apple said the Mac mini with the M6 chip delivers up to 40% faster CPU performance, up to 4x faster performance for AI tasks in particular, up to 2x faster graphics performance, and up to 2x faster storage speeds compared to the previous-generation model with the 10-core M4 chip, 32GB of unified memory, and 2TB of storage."...Show more →
With the release of the new iPhone 18 today, we get a bit more information on the M6 architecture as this iPhone is based on a similar layout. The benefits to Denoise times will be dependent on how LR takes advantage of this. Looking forward to see some results from the M6 Mini.
New neural accelerators built into the GPU offer twice the FP8 performance of the previous generation, which helps the chip run large language models locally on the device.
The A20 Pro features an additional Neural Engine, bringing the total core count to 32. The company says the redesigned Neural Engine delivers 2x the compute of its predecessor, is more efficient, and now supports 8-bit floating point math.