Pre-training equips text-to-image (T2I) models with broad world knowledge, but this alone is often insufficient to achieve high aesthetic quality and alignment. Consequently, supervised fine-tuning (SFT) is crucial for further refinement. However, its effectiveness highly depends on the quality of the fine-tuning dataset.Existing public SFT datasets frequently target narrow domains (e.g., anime or specific art styles), and the creation of high-quality, general-purpose SFT datasets remains a significant challenge.Current curation methods are often costly and struggle to identify truly impactful samples.This challenge is further complicated by the scarcity of public general-purpose datasets, as leading models often rely on large, proprietary, and poorly documented internal data, hindering broader research progress.This paper introduces a novel methodology for creating general-purpose SFT datasets by leveraging a pre-trained generative model as an estimator of high-impact training samples. We apply this methodology to construct and release Alchemist, a compact (3,350 samples) yet highly effective SFT dataset. Experiments demonstrate that Alchemist substantially improves the generative quality of five public T2I models while preserving diversity and style. Additionally, we release the fine-tuned models' weights to the public.

While deep neural networks excel at a variety of visual tasks, obtaining large quantities of labeled data remains exorbitantly expensive or time-consuming, especially when it comes to pairs of photos and their artistic representations. To overcome the burden of annotation, various solutions exploiting the unlabeled data have been proposed recently. In this paper, we present a novel approach to the unsupervised domain adaptation problem, allowing us to successfully generate avatars from photos. Assembling a system of several neural networks, including a Generative Adversarial Network (GAN), fully trained on unlabeled data, we researched the influence of various factors on the GAN training process and eventually built a system superior to current analogues. In contrast to the existing unsupervised domain adaption approaches, the proposed solution is highly flexible, which allows tuning of individual elements of the system to achieve different visual results. During the user study, we evaluated the performance of the proposed method and found results close to the human level quality.