271379 2024-03-21 07:22:11 -0700 [Nicosia] Images with translate-Z get unloaded when out of view 2024-10-30 13:45:52 -0700 1 1 1 Unclassified WebKit WebKitGTK WebKit Nightly Build PC Unspecified RESOLVED FIXED https://bugs.webkit.org/show_bug.cgi?id=255823 P2 Normal --- 1 kdwkleung cgarcia abrarsl2002 bugs-noreply cgarcia mcatanzaro nekohayo sidwirb oldest_to_newest 2022724 0 470457 kdwkleung 2024-03-21 07:22:11 -0700 Created attachment 470457 Blur test If translateZ() is used on an image, the image gets unloaded when scrolled out of view, and when you scroll back to the image there's nothing there. Reproducers: 1. flathub.org (open an app's entry, scroll the screenshots in the carousel in and out of view. The image is unloaded. Then in Inspector highlight the image and uncheck -webkit-transform: translateZ(0); The image pops back immediately. Now even if you scroll the screenshots in and out of view the images stay there.) 2. apple.com/mac-pro (scroll all the way to the Design section and activate the effect of taking off the case from the Mac Pro by scrolling past it. Then scroll back up. The case is gone. Now open Inspector, highlight the image of the case and uncheck z-index: 1; The case pops back immediately (but (correctly) behind the Mac Pro)) 3. I attached an HTML file that is used to test performance with blurs in another bug report. Open test-gpu.html. Scroll up and down repeatedly until all the blurred images get unloaded. Then reload the page and do not scroll yet. Open Inspector and uncheck all the transform: translateZ(0); Scroll up and down repeatedly. The images do not get unloaded. This can be observed in all WebKitGTK versions, whether Cairo or Skia. 2022725 1 kdwkleung 2024-03-21 07:22:48 -0700 This supersedes https://bugs.webkit.org/show_bug.cgi?id=255823 2022735 2 mcatanzaro 2024-03-21 07:40:29 -0700 *** Bug 255823 has been marked as a duplicate of this bug. *** 2022783 3 nekohayo 2024-03-21 09:26:50 -0700 Would this be the explanation for bug #250978, or that one is a different cause? 2030895 4 mcatanzaro 2024-04-24 09:42:27 -0700 *** Bug 250978 has been marked as a duplicate of this bug. *** 2030902 5 nekohayo 2024-04-24 10:00:34 -0700 Easier than https://apple.com/mac-pro as a test case is https://www.apple.com/ipad-keyboards/ ; if you scroll down the iPad keyboards page to the bottom and back up, all images are unloaded, which makes it easier to notice. 2030953 6 mcatanzaro 2024-04-24 13:03:03 -0700 I've never noticed this until a few weeks ago, so I mistakenly assumed it was a WebKitGTK regression. But I tested as far back as August and the bug still occurs. If it's a recent regression, it's not caused by a WebKit change. 2040552 7 abrarsl2002 2024-06-09 21:39:15 -0700 This can also be observed at https://www.codeedit.app/ 2040600 8 cgarcia 2024-06-10 06:22:01 -0700 Pull request: https://github.com/WebKit/WebKit/pull/29672 2040741 9 ews-feeder 2024-06-10 23:49:28 -0700 Committed 279898@main (3deb4217571c): <https://commits.webkit.org/279898@main> Reviewed commits have been landed. Closing PR #29672 and removing active labels. 2071700 10 aperez 2024-10-30 13:45:52 -0700 *** Bug 276076 has been marked as a duplicate of this bug. *** 470457 2024-03-21 07:22:11 -0700 2024-03-21 07:22:11 -0700 Blur test blur-filter(1).tar.bz2 application/x-bzip 4350 kdwkleung QlpoOTFBWSZTWb4Ga1MABb1/////////////////////////////////////////////4Au/fAH3 ztrvefWT2+3OcXffV8o6A3vpnutd8upXvd71Z69DUKNCaZMptMU9TepqmMjKbFPEyjwU8k9Nqnqe p6aNI/U2imaaGUeppmiabU2p6T1PKepp4jKbJlNNpPU/VNomngo8p6ZBlHqbU9NTwSGynqMMmJqH qbak9MUwEyPSPUIpoTRPRMU3qamRmobSep6hkNNMIwhpoBoxMah6Mp4oZNkTZTTBqbUaDGkBtQZA AaBoNBoaZAAANGhoGRoA0DEDQ0aaERICTwj1T0yankaNTZGINTyT00niRtTJhNGmIxHqaZGhoaaM TQaaAGgAGmjTT1ADRoaMmgwgGgHqMgeoANA0AGQNAAahJpk0p+RPU0NDSbQg0xNDJkaDQ0yaNGmQ MgZNGgAZDQMjQADT1AGg00aaAANGgGg0AA0AAAAAANNADTRIRDFT1P1NMSfqn4go21TT2kmntU9Q 9ENPaFNpogyG0T0TTIaDTQB6mjR6jQaNNqDQAaA0yBtQBiAANGmgNB6hkyA0DQYhoASpqIjRT9PS GjUzUynqeJNknjTUTyBqbTU9Gp4o9Ieao9Go008jSPKHqemp6mm1NGxRnqjZpTygGjymZE9CBk0D 1M1DaCeo2U9R6g9TRtIAGQ0yaDRp6gUvZ2LdmItO5BRpdR1T+1YIJYFzSgw87RbTGNVZvNTJNo/v wkCCIt9JUAsiAcWRG7vkc/p+ffRNUhRP7IAnFESRTKKhMumPSqc8Utyk4/uQE0RAS8NH3ihBkSEo CkLp+4ZABRXwK63K5Wkc8SCgwZQKCDbKgk4atnvh5P5jKiU5cRrQBwlnDr3gHgaH6V1MPVBkAJRV RJYrjRvcMALlMJkQrQDDufYA4/0lTXA4coJh2uZUJZAZ7ovvcP+LFNy8oegwUoN3IooKksEBKU1c JAYFGJBFKyYpJb6tWM4PhBBVgvwDFQ30YQKHsD8d68m68PgtHdvRaChQk8xLi4S0f5eCpBlKo7qW olITSKlaIUuFBA/R3S6GIM0ykycbuzrDGrnDY7Bm4WPXys3Jv88iETKkU+GdntuSCQ4U1MVBIyQK jBK65vwmvuevx1xob6YrC4MZjMr2YUETOIAPDVRc6pLkkeH1Tj1SJhiUAyEEE3ysEFn6iVS1JqGP qDBNMGBlsA1sa8cmfKWKZ2twI3wO8xiGBcABG7GASAqBUXKhUM7CotS5eGIDAVplhBy93o6DsxIT obNxVKHZ3tEzaEB71YIZ4Dof4qAhJzmlMgVxpIHgfDbe2IPKQkivBpw6NAiGC7wWWeCIiXFz6GEv uuojl9ZiNW1C4vXsxD51LQ74Gs+/bCwQMuDA4JE/ahIdLSLxDSZLE/J58InEUb3xOCBhqKmHM+/M iZ6TO0h1WYPTiW8rJXHM4IoaAGPoin/+5+LYog3oxZcYS83VFhRoxIQG7x98Pm2l6f+9ZnhiBORC SdWYWwIPU4FAg7N8b40RBhENcZde0PBo22HkwPFbr73kYvsV5IRim2pNYNC1LD5GcZUE8GcMn2dV dpUTVHB8mvKkv0/yh7Dmtq9Il5e1/KH/ZF2qN9XCNDIQOo16udaJBkbDeegf1snvcyoIG0Au/ZmA EONlSZUrZt73+6AKsAtgC9AXAp5lN5HDrsG95RUOY3zNehe3PTYFNoW3huDHACBTY0IUTkAzB6aY uJzNXvgA9qepUUwrj2Ny9dTWukdPTZjvh2TxhXCpQDNd3o3I9NCwjapv1pUUIK7zYPTKjm5GwdYw 9JFvous08ruzM50nFrnNLREBevAKIAuvR5+4oqo5VIeKBRoilIomXls/bxnmxKSSwPvnON0eSa3f hmYiJbaKObws+PX1aip0sSsoVcPe6aojrE6UUyTUTTdXdFsI7VLZjjeQAV2QFvijpM03B7p5+8vJ IYJAAs3qfUPKo8MBUQwA7Z6hCvkyaUyz+JNgfnq/NAXAIz4yzmAUyAQEAoDXqGvsNWBJpbAAGfSb l9xu94qyxoHdtk8rCgQuxvtpntlzUTDKcgMuAKEFDACJcWMwocT6FFVOBtoaSYB3aUW3NiO1FuSo dHEWoqNcTHCVVE56ABSDwGoE8G3qrMIgcF6XQy1dol/qnDs9bB2TvUtxNHw+PZIHhmMSAAKCHGYv MenmmlwxPVfBplAwUAAOK/UOQAVGFrDQACtB2c4qBCygAU52ZEmc1O4mWZyjWjXplKAF8o0e6tV8 5SPvUokp5rlA6iJPNN8ergGGBOz7LRgKd+rsIPZxzW/AEfKU7RSkt+vtv1fgOWnBQrACJAMvrHNd JRQBv4xQGNuY3/HEt+T04KK0yUFl5ERlp6paLsEs0tNKJNP9WLsuLYOYudmY2u0dAh6PlYvte5s3 ijIuCY6gQ1rGtoACn/NFIQA74hFIbq90T0NFMob+1jbMVDOEcvTtZY+wZZ591cqotm21KCFKjfTG C6ZDwEi5NVR/noSLo0oYGroV1ONk56c25BXQO585MCmMRDqZPL51h6rQ6Z0112tL8GkW3jpyOfqI aBCgFEvaJbP7DUmtcQBxYf6cbf2eEMgkbfhmiZSGtLUylbD5A/Jds7TkBgTCdrGdagUMYKpJOEh7 TzR8ojUOglxDBdcKBBx6ziJmUK+uAcIamllB0L+hXx8KqkbBrll/QdSABUUl0bqaDDYogAQZf+fT MF/Ms6/+j5qF8O0FlSvoxX3n2CL2vVmuKYc0qBSH56lSWKJmdCwiLiosFNRGcCrrJD9Ch2SD/M4K PHeWjvf8jTDZQ4852X8+499w6nOrzN5vbbPqF9eJctpDopy2t2cdZ1FpG/G2rkFBM568w0KiblGs JKPTQh2eVu6wdOac96RDtO6MAFa0JfVaZhsYYYaLGAH0AZTx6zeLSln1yJLuscwslUWxDOVUnlNQ QMYRaYnr2hbOHH/XR1eb3EKwrWaV6g/zsgZC9AUBn9+V0GgRWwoNI9mlUd+zYRtBc56/yE8IkWid 0ID/4Nc6AwK28UBnwDSBoAMQDOcWsKhFFcSF0WkvNFb3exJ/jtbWwsmFtGAXc7vby1ANbk9+49Qy WZwYdUn7K5tEfmfNgFw6AEYAQAAlAtjZU/2RGR6t7u3lW3coTR6ITTSZoM2CCo1zCXvTBpFEon4Y CLAOBtY5g5EyBCoCitEVcsVEDhhAY+W2d+uvKIp+Ap7JJ78ZQK8AyOXAacCb1yKyyFGm47yuc9t6 38PABYAoV8pDUFWBMqAx0VI6mn4mhfYXOGrsJHiYvhQgEeCjbyABPAHCluQgmikbmdYy138yogyt HoGwnd3HmbJZtMtrpx6SbuQAb2j6vtEBY4DSrlhEEE2tklggidygSM4tGXawwjRBfXYbY3fQAZDF hgCm1IKGAURcr0N3kqgB/UAec8GXS5UuIP7MK3Xf8HYXzPrspIz4BFxwPEVH1ztdkig6dIO1fVre MAdbKBgiUdNuVG/pJVKgI2P19SvaAC9I3waTq+38dUxG/tHg9BQ6B+QAlwDet2ogBDgDQDugNJAv dvgvIMsW7GnhCsAIqzAFyCBYAkNFw+PuKows44Hzkq0DHgQoKGp/PGLqHhQBoAUn/c85mOE2hC+g Nt76KlW4KIzlgyN4zWPMTWopUuim3Ji/AnGX32fJ/vnaW9qHLkAyMrY7a/WZZaJxcXQiBAABOHSZ j2gHENlyIScBQcWYsOQOWEhXnNMStQ8o7xJ1D4F90le4uMGip3E+cKpkuVm27SLV/SSSd9T1trgo bK3m0sNsHbaqXFFINm0kaqL3Ye/7xm/d5vlyittZWfA8JaZyxwmr6KafFCUUOWXQqKWW1Nskj+wN eNeMKzJEOg1zJ4r89fEgCl7msJUtFbGBzSIE3fFTK8dxRATZdwaIQf7OfJB9pwNbn1TSCWq0/u0k J0HO7ZTcOMLG3TH4cn7uJCKP+l5Nx76rblGf0JivxR128xto4InWItli4iiBtWwizC6sauM6eZii rGl7eUini0wjq1kO/Oxl7fCxbOAOrMxBJmyHchgSPJwgX8IP38KNJej1cXW1HdMgL6EaPom+xX95 O1ozq5Wu1S/9o7fYhTXYjpSbR4YqVmO3f+ZZGEcKZWZj3KDQZdfg4l5awy3rens25jbzBs/AcZIi RWAy8zQPMXEwKJzNkVCUOgAXP59yalRlgJWRDGyTZXEJSRhMD+kq/PaJ1kIqaKEDASd9dKqwZNrW HfW8T4SC5h+mrhesVGOfdjuF4u/FL9PRy6ibat6pw71sbouegF85Kh73J0kgzX3xrKJl3xl68nnt ujQlNNvqYlhv6fGJ2+zyTnSRFKIyfHI+ailIqVVU90YnIQwFlOLfDj2Sw9exKTkxyelEjSM5G/4f XGj0n6NPMhLq/y+xlkcWmsctEQgG5KBTXMh3j5p0RghCC2cNSK7nWpkGNht6BA8UhXqHvv1Tnppj nJkhUraNcQsUuZbPYxW7WOsQ2K0OMk0sOAE0W0Zkh6/Jdpg5nswJH0HVSXAaf5b1uPq+d2G1yzkq Z6jkBmzqClrmfFhy3ntIt73qyH+MlVd/q1ABIEAy0VM7mOOGZX+zBP1ZnGx48yrxEKE45z4EE4Ig un6v9MTueMJljdzoyuE5cxlB18VHKJVepinVOi0E4VwfZxkIXQQAmFLKlrgrykPFh5DhZGJVHZ8D BABoQLBwNWmEmpHpHw2dk+/bunKaPCcEZlJAlAWgpUdBLrAdkW60gpaeFuHzg6dw6oaTIow3MX4u 0tRwUGyPmy3A9M0Usy0KF2hxwnBAFMfIA0SMmEMBDGLfunlaV2AcW1FQxx8UrnFqBjXLu/IPsjA9 g+5v71Jm6s8PvW11IhCZQq+NCzaVtfc0YQf0RRhNFjYmatLTMRvmbuwTS7g3EZVc78UQmBBkjqv7 csnkOqTtEmtl8LSpRC5hhQGwDAmkWaAq0QWvtnpmPUimYYiDp77MSXUwKTXr6K/UTdx2pk/WcOfd glk9I/xiRGrI/jQ2zNzPuIMDdRZkQTFAlcRLEqtqMs8uACBujyeC8G21hgUcrkztMirhhCvZgMAU Ozrcecqe56ZqjIopvDTYO8/vq7uljNbdF2FKSwyBhgcgwHYqd9RN6eLvPBKR9HIT3Dhdsios6DXn JjvODAVZnb+W9NUeWov9a2XddXOuLBd9vBvnz5PE6+1Gfm8Rb6CzeCc0/x2xyHS4h4IpzVIpCAkB vEbOiKNj22kZpzjXJA+boNtWOLMLwu5aM+gKAYIVRG1RIWDpWe5VMNgD1OIz8034tdpgRuI5bEK1 hWK08FTkKpUdMV4p4S5StupsB3rq9LoYj+AgsR9nAjmm/qpYlgp8eTYmLZeyL41jrZGwea7hAqoJ JelO9cKHJpBb7YtNREW2af7UngfCkgHy0yhx1rJYW9szMGzeq2DWP3OcSz+xYKH+489bubmLnzx/ SMb9stCOtkxQbEs4UYYKQpQAUFienhMEjzGMlPUO5rx16zXquWvci8At2svVsyYY3hn5V5WnIRWV /qS/R4SUmRewmEb8pka/cXq+L64dN128k8IgUKa1obhHombNOdGwuPSSCxkpXfNRLwpCbbA7KDkP WTPIXvDaL4CPSUsNuYDptd/Humcx6y6dbZjNo4C4kTPUOyna7LJwD/FWh0qpKC75afvrkhe7TJt8 dU1cp5aVaAKWYVhUD1JDlvVyc1VsQcup5nnpodFqYbsFetZCO7PFnX5VlxnhODeWmvKw1LgIMrHy KJt8U9BSf+LuSKcKEhfAzWpg