Fix validation errors by adding the STORAGE and SAMPLED usages to any image that needs its mips computed

Also stop computing mips for images that have some regions specified already in non-base mip levels

Author: devshgraphicsprogramming

examples_tests

@@ -847,12 +847,17 @@ auto IGPUObjectFromAssetConverter::create(const asset::ICPUImage** const _begin,
         }
     }
 
-    auto needToCompMipsForThisImg = [](const asset::ICPUImage* img) -> bool {
-        if (img->getRegions().size() == 0u)
+    auto needToCompMipsForThisImg = [](const asset::ICPUImage* img) -> bool
+    {
+        if (img->getRegions().empty())
             return false;
         auto format = img->getCreationParameters().format;
         if (asset::isIntegerFormat(format) || asset::isBlockCompressionFormat(format))
             return false;
+        // its enough to define a single mipmap region above the base level to prevent automatic computation
+        for (auto& region : img->getRegions())
+        if (region.imageSubresource.mipLevel)
+            return false;
         return true;
     };
 
@@ -1025,8 +1030,8 @@ auto IGPUObjectFromAssetConverter::create(const asset::ICPUImage** const _begin,
         promotionRequest.originalFormat = params.format;
         promotionRequest.usages = {};
 
-        const bool integerFmt = asset::isIntegerFormat(params.format);
-        if (!integerFmt)
+        // override the mip-count if its not an integer format and there was no mip-pyramid specified 
+        if (params.mipLevels==1u && !asset::isIntegerFormat(params.format))
             params.mipLevels = 1u + static_cast<uint32_t>(std::log2(static_cast<float>(core::max<uint32_t>(core::max<uint32_t>(params.extent.width, params.extent.height), params.extent.depth))));
 
         if (cpuimg->getRegions().size())
@@ -1036,7 +1041,10 @@ auto IGPUObjectFromAssetConverter::create(const asset::ICPUImage** const _begin,
 
         if (needToCompMipsForThisImg(cpuimg))
         {
-            params.usage |= asset::IImage::EUF_TRANSFER_SRC_BIT;
+            params.usage |= asset::IImage::EUF_TRANSFER_SRC_BIT; // this is for blit
+            // I'm already adding usage flags for mip-mapping compute shader
+            params.usage |= asset::IImage::EUF_SAMPLED_BIT; // to read source mips
+            params.usage |= asset::IImage::EUF_STORAGE_BIT; // to write dst mips
             // TODO: will change when we do the blit on compute shader.
             promotionRequest.usages.blitDst = true;
             promotionRequest.usages.blitSrc = true;
@@ -1095,12 +1103,8 @@ auto IGPUObjectFromAssetConverter::create(const asset::ICPUImage** const _begin,
 
             if (needToCompMipsForThisImg(cpuimg))
             {
-                // Todo(achal): Remove this API check once OpenGL(ES) does its format usage reporting correctly
-                if (_params.device->getAPIType() == EAT_VULKAN)
-                {
-                    assert(_params.device->getPhysicalDevice()->getImageFormatUsagesOptimalTiling()[gpuimg->getCreationParameters().format].sampledImage);
-                    assert(asset::isFloatingPointFormat(gpuimg->getCreationParameters().format) || asset::isNormalizedFormat(gpuimg->getCreationParameters().format)); // // for blits, can lift are polyphase compute
-                }
+                assert(_params.device->getPhysicalDevice()->getImageFormatUsagesOptimalTiling()[gpuimg->getCreationParameters().format].sampledImage);
+                assert(asset::isFloatingPointFormat(gpuimg->getCreationParameters().format) || asset::isNormalizedFormat(gpuimg->getCreationParameters().format));
                 cmdComputeMip(cpuimg, gpuimg, newLayout);
             }
             else