[Sony/罪大滔天]相機集中討論區 (143) 買左85GM II 未？

自己反而覺得會繼續出
佢一塊cmos可以重複咁用咁多次
就算講近期d新機既
apsc果塊新cmos都用左fx30 zve10m2 6700 3部到

要又平又wide又大光圈有唯卓16mm 1.8

都係siii d料仲要全部都係同其他機共用

用開mac機執
執完直接up入相簿或者直接airdrop

嗰三款機完全唔同款唔同特性食唔同客源喎
你攞返同一塊sensor dum落另一部a7s啲人點會買

你係問電腦操作定係點揀格式

咪掉落icloud
win機一樣得

冇錢買Mac機，
冇用開

睇你用邊邊就手
app丟落folder就自動sync
web就自己揀嚟up

再唔係你鍾意用lr adobe creative cloud又得

MacBook Air m1家下應該好撚平
btw 印象中window都好似有得裝icloud
執完直接拖返落去icloud個file入面就得

啱啱入咗m3 mba唔用啲AI功能都算順
平時主力用desktop執 mba就留返旅行

雖然係，但最好等多幾日睇埋M4先，始終粒NPU快左近4倍，做AI denoise應該會快好多。

最多買MacMini

好唔實用....又放唔到其他野

用ai既 我M2 Max mbp 32gb都唔係幾掂
r機d相做denoise 如果一多都係要慢慢等

我M2 Max mbp都等緊m4
不過唔急無太逼切性 慢慢觀望下先

4倍係同m2比m3同m4 gpu好似冇乜分別都係10粒

好似好掂咁... 不過點解唔加埋2470GM2一齊比較

睇錯gpu 自膠

Here is a summary of the NPU INT8 performance (in TOPS) across Apple's Silicon series, from M1 to M4:

- **Apple M1**: 11 TOPS, using a 16-core Neural Engine.
- **Apple M1 Pro/Max**: 11 TOPS, 16-core Neural Engine.
- **Apple M1 Ultra**: 22 TOPS, with a 32-core Neural Engine.
- **Apple M2**: 15.8 TOPS, 16-core Neural Engine.
- **Apple M2 Ultra**: 31.6 TOPS, with a 32-core Neural Engine.
- **Apple M3**: 18 TOPS, using a 16-core Neural Engine.
- **Apple M4**: 38 TOPS, also with a 16-core Neural Engine, more than doubling the performance from the M3