作者: chimp 時間: 2015-3-31 11:22 標題: [壓縮PNG]有冇人睇得出壓咗有唔同?
提示: 作者被禁止或刪除 內容自動屏蔽
作者: bennetng 時間: 2015-3-31 12:19
就咁 printscreen 佢網站, 然後將幅 printscreen 圖掟俾佢壓縮已經睇到有唔同[attach]1777422[/attach]
作者: minazuki 時間: 2015-3-31 15:49
Advanced lossy compression
本post已完
作者: gamezz 時間: 2015-3-31 22:15
回復 1 #chimp
右邊隻腳
via HKEPC Reader for Android
作者: dvd-rw 時間: 2015-3-31 22:44
FILE SIZE爭幾多先?
作者: 高登阿伯 時間: 2015-3-31 23:43
回覆 1# chimp
網頁大大隻字Advanced lossy (有損)compression，壓咗當然唔同喇。
作者: chimp 時間: 2015-4-1 01:26
提示: 作者被禁止或刪除 內容自動屏蔽
作者: hihihi123hk 時間: 2015-4-1 01:58
本帖最後由 hihihi123hk 於 2015-4-1 02:02 編輯
網上上傳圖片，再下載，其實唔關程式編寫事
有人話睇到壓咗有唔同，叫我唔好用。想求證。我的相好似header ...
chimp 發表於 2015-3-31 11:22
識用一定係用呢隻 http://pngquant.org/
既CLI 自己慢慢tune 
Algorithm
pngquant uses modified version of Median Cut quantization algorithm and additional techniques to mitigate deficiencies of Median Cut.
Instead of splitting boxes with largest volume or number of colors, boxes are selected to minimize variance from their median value.
Histogram is built with addition of a basic perception model, which gives less weight to noisy areas of the image.
To improve color further, histogram is adjusted in a process similar to gradient descent (Median Cut is repeated many times with more weight on poorly represented colors).
Finally, colors are corrected using Voronoi iteration (K-means), which guarantees locally optimal palette.
pngquant works in premultiplied alpha color space to give less weight to transparent colors.
When remapping, error diffusion is applied only to areas where several neighboring pixels quantize to the same value, and which are not edges. This avoids adding noise to areas which have high visual quality without dithering.
作者: henrywho 時間: 2015-4-1 10:51
識用一定係用呢隻 既CLI 自己慢慢tune
hihihi123hk 發表於 2015-4-1 01:58
ching 有冇使用心得?
作者: gamezz 時間: 2015-4-1 12:47
回覆 8# hihihi123hk
I am sure not much people have time to tuning the image file in real case.
作者: 3ldk 時間: 2015-4-2 00:53
識用一定唔會用依幾款有損既壓縮software, 無謂用有損既programe污染左張圖片
http://pointlessramblings.com/po ... s_optipng_vs_pngnq/
不過佢個測試好奇怪, 我明明記得optipng係無損的, 點解個測試係有損? (另外三款我確定左係有損, 一定唔會用)
選得png就係唔想有loss, 所以我一定用PngOptimizer
因為夠簡單有GUI, 又support unicode, 又可以係合理時間內優化到合理size (因為我以前曾經用佢比較過另一款打command優化的軟件, 用超高優化參數, 花多5-10倍既壓縮時間, 但同PngOptimizer比又唔係細好多, 所以我而家追求合理就算)
作者: hihihi123hk 時間: 2015-4-2 11:02
回覆 11# 3ldk
如果要黎儲 source緊係唔會用 lossy
但係如果做 web / app , 用家擔心 速度多過你 png 係用 lossy/lossless ， 同埋 如果租 Cloud Service 既，每個流量都係錢，壓到幾多% 就慳到幾多%既錢。
我試過designer AI export highest quality 比我
2.1 MB ，再壓到 200KB, 正當size 望，根本睇唔出分別
作者: hihihi123hk 時間: 2015-4-2 11:16
回覆 hihihi123hk
I am sure not much people have time to tuning the image file in real case.
gamezz 發表於 2015-4-1 12:47
Yes, I think there are no ppl will use RAW-> png to store the image. No one wants to ruin the source.
using png because it support transparency for image merging/ overlapping .
If you have lossless source yourself then keep it. if you want to publish it on your server , I will suggest to have a compression.
tune to have an acceptable compression to be your standard.
"tune once, use forever"
make your web/app load faster and save the traffic cost
作者: 3ldk 時間: 2015-4-2 15:30
回覆 3ldk
如果要黎儲 source緊係唔會用 lossy
但係如果做 web / app , 用家擔心 速度多過你 png 係用 ...
hihihi123hk 發表於 2015-4-2 11:02
其實只係我個人認為公德心問題
我覺得用png就唔好用有損
要有損, 要慳size, 不如用jpg, 而家D jpg, 一樣可以2.1MB壓到<=200K, 而且一樣睇唔出大分別
但如果用jpg, 就可以比人知道依張係有損, 用png就一定會誤導人
咁等同有D人唔識電腦, 用png壓完jpg再壓返做png一樣道理
可能你覺得D素材自己畫, 自己有個source就得, 但一個人無可能乜都做得晒, 點都有D係網上download落黎, 如果個個都將D唔知lossy定lossless既野流出去, 咁以後download一張png, 都仲要花精神去研究係真定假, 好似去淘寶買野咁
作者: blue01299 時間: 2015-4-2 17:27
回覆 14# 3ldk
JPG冇透明
作者: minazuki 時間: 2015-4-2 18:46
用得壓縮PNG係貪佢有半透明
JPG對線條分明處理係勁差，用256 GIF/PNG都好過佢(只係講緊線條分明)
而壓縮PNG就係集合左256 GIF/PNG既size優點 + 半透明
作者: henrywho 時間: 2015-4-2 19:40
其實只係我個人認為公德心問題
我覺得用png就唔好用有損
3ldk 發表於 2015-4-2 15:30
其實 resize 已經係 "有損" 啦 ....