Mirrors/overte
3
0
Fork 0
mirror of https://github.com/overte-org/overte.git synced 2025-04-05 21:12:25 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions 2

Restore KTX version to 2

Browse source 
This fixes compatibility with older baked assets
This commit is contained in:
Dale Glass 2022-12-02 01:54:42 +01:00 committed by HifiExperiments
parent 1c06c8652c
commit 101592357f
1 changed files with 23 additions and 7 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

30
libraries/gpu/src/gpu/Texture_ktx.cpp
View file

@ -28,12 +28,25 @@ using KtxStorage = Texture::KtxStorage;
std::vector<std::pair<std::shared_ptr<storage::FileStorage>, std::shared_ptr<std::mutex>>> KtxStorage::_cachedKtxFiles;
std::mutex KtxStorage::_cachedKtxFilesMutex;
/**
* @brief Payload for a KTX (texture)
*
* This contains a ready to use texture. This is both used for the local cache, and for baked textures.
*
* @note The usage for textures means breaking compatibility is a bad idea, and that the implementation
* should just keep on adding extra data at the bottom of the structure, and remain able to read old
* formats. In fact, version 1 KTX can be found in older baked assets.
*/
struct GPUKTXPayload {
using Version = uint8;
static const std::string KEY;
static const Version CURRENT_VERSION { 3 };
static const size_t SIZE { sizeof(Version) + sizeof(Sampler::Desc) + sizeof(uint64_t) + sizeof(TextureUsageType) + sizeof(glm::ivec2) };
static const Version CURRENT_VERSION { 2 };
static const size_t PADDING { 2 };
static const size_t SIZE { sizeof(Version) + sizeof(Sampler::Desc) + sizeof(uint32_t) + sizeof(TextureUsageType) + sizeof(glm::ivec2) + PADDING };
static_assert(GPUKTXPayload::SIZE == 44, "Packing size may differ between platforms");
Sampler::Desc _samplerDesc;
Texture::Usage _usage;
@ -43,13 +56,14 @@ struct GPUKTXPayload {
void serialize(DataSerializer &ser) {
ser << CURRENT_VERSION;
ser.addPadding(1);
ser << _samplerDesc;
uint64_t usageData = _usage._flags.to_ulong();
uint32_t usageData = (uint32_t)_usage._flags.to_ulong();
ser << usageData;
ser << ((uint8_t)_usageType);
ser.addPadding(1);
ser << _originalSize;
@ -60,15 +74,16 @@ struct GPUKTXPayload {
bool unserialize(DataDeserializer &dsr) {
Version version = 0;
uint64_t usageData = 0;
uint32_t usageData = 0;
uint8_t usagetype = 0;
dsr >> version;
dsr.skipPadding(1);
if (version != CURRENT_VERSION) {
if (version > CURRENT_VERSION) {
// If we try to load a version that we don't know how to parse,
// it will render incorrectly
qCWarning(gpulogging) << "KTX version" << version << "is different than our own," << CURRENT_VERSION;
qCWarning(gpulogging) << "KTX version" << version << "is newer than our own," << CURRENT_VERSION;
qCWarning(gpulogging) << dsr;
return false;
}
@ -79,6 +94,7 @@ struct GPUKTXPayload {
_usage._flags = gpu::Texture::Usage::Flags(usageData);
dsr >> usagetype;
dsr.skipPadding(1);
_usageType = (TextureUsageType)usagetype;
if (version >= 2) {