基本上參照 AGC91V10-T2 內容 fully copy, 我只試拍幾張看大致沒差異太大~
從 AGC91V10 開始, 它的 libpatcher 分 global patcher& individual len profile patcher, 簡單說:
- Global Patcher: 所有 len/proifle 參數有共同設置值, 只需要設定一次即可
- Individual len profile patcher: 獨一無二的值, 如 1X/0.6X/3X/10X luma/spatial/temporal不可能都相同, 像這樣設定就得一一設定
Global Patcher 可以大大縮短配置檔設定時間
我已上傳3個 package naming APK 檔案與 T1 配置檔, 可上去下載安裝試拍測試看看 ~
P.S.
這次我沒充分做測試不知道結過是否與之前相當...希望是... 但若不是會持續改善
但每次的調配過程若有收穫是最好的經驗累積與學習
AGC 9.1 V11.0
- Fixed the patch invalidation problem caused by some parameter conflicts. All V10 users please upgrade to V11
- Fixed Global custom patcher
- Fixed full device crash by default
- Fixed sect、tone、gamma not working
- Added Link HDR Mode: ZSL、HDRE、HDR Net
- Turn on Video Stabilization for pixel device by default
- Removed parameters: ShotParams Layla, Oriana, Eevee, PortraitRequest Eevee
基本上相容於 AGC91V10 配置檔, 但近日會搞一隻 for AGC91V11 配置檔, 優化 denoise 組合, 測試完會 release
P.S.
AGC91V11 配置檔一樣會使用 global patch 去縮短將來配置時間 ~
"
So in this case, 'a' would be like the gain? (Maybe I'm getting bad translation on my part but 'a' translated as velocity which I'm interpreting as gain here). 'B' the main reduction?
That seems odd, since, for instance, we can set B to negative values and it works just fine (but differently). Set Luma 'a' to a negative value though and all noise reduction turns off. Luma denoise, luma noise, etc, none of them work anymore."
"如果按照RawTherapee的pc版使用体验来表述的话，一个设定的A表示力度，B表示对A这个力度的还原纠正"
"B为负值时很好，这更加表明它是一种还原纠正。举例当A值固定时假如B＝0为不还原，那么当B＝1就还原一定的力度（还原了就会产生一定噪声），而B值越小就越不会还原也就越少的噪点，所以降噪效果就更好。"
"
Let's say I set both luma a and b to 0.125 (the lowest it will go in this mod before going into negative values). Now if I increase b, it should restore the image somewhat. But it doesn't, at least not wholly. Mostly it just further reduces the noise.
However, there are some values (on the way to 1 from 0.125 let's say) where it changes the shape of the noise, makes it harsher, then softer, but overall increases the noise reduction, if only slightly. It does this in the opposite direction as well, if one were to take 'b' into the negatives.
"
挺有趣的, 關於 Main Setting A/B 設定值另一解釋, 經過消化後, 理解為:
1.A是增益, 意思就是denoise 力度;
2.B是增加或減少 noise, 越往正值, 增加 noise, 越往負值, 減少 noise.
對多測試才知...
AGC91V11-T1 is for "AGC9.1.24_V11.0_xxx.apk", xxx=aweme/samsung/ruler
AGC92V3-T1 is for " AGC9.2.14_V3.0_xxx.apk", xxx=awesome/samsung/ruler
兩個配置檔基本上內容一模一樣, 可以互用, 均使用上 Global Patch Settings + Individual len settings.
基本上均在改善 noise&details.
有我連結朋友,記得更新
AGC91V11-T1 Test Shot
AGC92V3-T1 Test Shot
國外網友實際測試, 結論是: S24U 10X 比 S23U 差 ! (打臉這篇幫 S24U 宣傳(https://hk.news.yahoo.com/samsung-galaxy-s24-ultra-reportedly-shoots-better-10x-photos-than-s23-ultra-113408159.html)
但若仔細看, 相差非常細微 ~~~ 一般消費者很難分辨出來
以常理而言, 光學變焦絕對比數位變焦相片品質好!
S24U 10x 是由 5x 焦距潛望長焦攝像模組放大 2x 得來，雖然感光元件升級但還是彌補不了數位變焦比光學變焦差的事實, 啥 AI cropping 是噱頭, 吸引傻B的消費者上鉤 ~
內文搜尋
從 APP 打開
X